CN107066995A - A kind of remote sensing images Bridges Detection based on convolutional neural networks - Google Patents

A kind of remote sensing images Bridges Detection based on convolutional neural networks Download PDF

Info

Publication number
CN107066995A
CN107066995A CN201710380211.6A CN201710380211A CN107066995A CN 107066995 A CN107066995 A CN 107066995A CN 201710380211 A CN201710380211 A CN 201710380211A CN 107066995 A CN107066995 A CN 107066995A
Authority
CN
China
Prior art keywords
remote sensing
picture
sensing images
bridge
msub
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201710380211.6A
Other languages
Chinese (zh)
Inventor
刘兵
周勇
郑成浩
王重秋
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China University of Mining and Technology CUMT
Original Assignee
China University of Mining and Technology CUMT
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by China University of Mining and Technology CUMT filed Critical China University of Mining and Technology CUMT
Priority to CN201710380211.6A priority Critical patent/CN107066995A/en
Priority to PCT/CN2017/089134 priority patent/WO2018214195A1/en
Publication of CN107066995A publication Critical patent/CN107066995A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/10—Terrestrial scenes
    • G06V20/13—Satellite images
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Remote Sensing (AREA)
  • Astronomy & Astrophysics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biophysics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Evolutionary Biology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Image Analysis (AREA)
  • Image Processing (AREA)

Abstract

The invention discloses a kind of remote sensing images Bridges Detection based on convolutional neural networks, for data volume and picture size all larger remote sensing images, detection efficiency is carried out to wherein Bridge position using conventional method low, of long duration.The present invention has initially set up convolutional neural networks model, and the bridge image that interception size size is w*h in remote sensing images initializes parameters in convolutional neural networks model, training sample is input in model and is trained as training sample.The remote sensing images to be detected window of w*h sizes is scanned according to step-length l in detection process, candidate window is drawn and marks good position information, finally candidate window is put into after model and exports remote sensing images Bridge position to be detected, realize detection.Feature extraction of the present invention without carrying out bridge picture in advance, simplifies detecting step, and the detection speed of remote sensing images is greatly accelerated while high detection rate is kept.

Description

A kind of remote sensing images Bridges Detection based on convolutional neural networks
Technical field
The present invention is applied to field of image recognition, is detected mainly for the bridge in remote sensing images, is a kind of base In the remote sensing images Bridges Detection of convolutional neural networks.
Background technology
Remote sensing image processing includes acquisition, denoising, enhancing, recovery, compression, segmentation, expression and description, the mesh of remote sensing images Mark detection etc..Wherein, target detection as remote sensing image processing a pith, in military field and civil area all Have great importance.In military field, it is necessary to carry out military surveillance to enemy and one's own side is monitored.By to defending The remote sensing images that star, aviation or aerospace craft are obtained carry out target identification, can understand landform, the dress in captured area The information such as standby, troop movements' situation.The Remote Sensing Target detection of early stage, which is used, manually to be carried out, but is due to usual acquisition Remote sensing image data amount it is very big, if carrying out interpretation using artificial, need repeated work, waste time and energy, and in real time Property is poor.Modern high-tech war, battlefield situation is fast changing, if image processing speed is too slow, it is impossible to obtain in time Key message is taken, causes to bungle the chance of winning a battle, one's own side is sustained a great loss.Therefore, remote sensing is carried out using fast automatic identification technology Image Automatic Target detection is extremely important to modern war.In addition to military important value, Remote Sensing Target detection The civil areas such as the condition of a disaster assessment of foundation and renewal, natural calamity in other side such as urban planning, geographical data bank also have And be widely applied.As the concepts such as global positioning system, GIS-Geographic Information System, digital earth system are proposed in succession, also get over More to need to carry out the target in remote sensing images accurate detection positioning.In addition, Remote Sensing Target detection is in accurate Drawing Also become to compel to be essential in the damage condition detection that two-dimensionally or three-dimensionally figure, natural calamity are caused and the change of target detection in city Will.
At present, the Bridge detection for remote sensing images, which is mainly used, extracts candidate region using conspicuousness method and carries Feature is taken, feature is carried out using grader to judge to obtain testing result.Patent No. CN200810232213.1 remote sensing figure As Bridge detection is to be trained modeling by watershed feature, remote sensing images Watershed segmentation is carried out with this, for splitting Result need during bridge machinery, detection bridge for the different template of different Bridge Designs, then extract Feature finally completes bridge machinery.
Based on the studies above present situation, the target of remote sensing images is primarily present following two problems：First, after pretreatment, The extraction of the artificial default specific features such as shape, length-width ratio or area of connected region is often carried out to sample image, so not It can guarantee that and extract effective or important feature, artificial experience influence is too big, and practical application effect is not good；Second, in order to not The minutia of image is lost, the process of artificial default feature extraction is also ignored sometimes, directly makees all pixels in image It is characterized, then these features is so done too cumbersome, can bring substantial amounts of superfluous as classifier training and the Back ground Information of classification Remaining information so that detection efficiency is reduced.
The content of the invention
Goal of the invention：It is an object of the invention to the advantage using convolutional neural networks in terms of image procossing, it is proposed that A kind of detection method that remote sensing images Bridge image is solved using convolutional neural networks.It the method overcome conventional method effect The low shortcoming of rate, by the feature in convolutional neural networks automatic mining image, finally realizes the detection of bridge picture.
Technical scheme：
A kind of remote sensing images Bridges Detection based on convolutional neural networks, including step：
S1：Training sample is gathered and pretreatment；
S1-1：The remote sensing images for including bridge area are chosen, manual interception size size is w*h sizes on remote sensing images Bridge picture；
S1-2：The region of bridge is not included on remote sensing images, interception size size is w*h picture, is used as detector Negative sample be trained；
S1-3：The positive negative sample obtained in selecting step S1-1, S1-2, on the premise of picture w*h sizes are kept, Align negative sample picture and carry out flip horizontal, change of scale, translation transformation, rotation transformation and whitening operation；
S2：Convolutional neural networks training pattern is set up, detector is obtained；
S2-1：Convolutional neural networks model is set up, and the parameters in convolutional neural networks model are initialized；
S2-2：The positive negative sample that step S1-1, S1-2 is obtained is put into the convolutional neural networks model that S2-1 is obtained, and carries out Repetitive exercise；
S3：Detect the pretreatment of sample：
Remote sensing image to be detected is chosen, is scanned, is laterally swept since the upper left corner of remote sensing image by w*h size windows Step-length is retouched for w/2, when the low order end of scanning to picture to be detected, a line is moved down according to longitudinal scanning step-length h/2, then from Far Left starts the step scan according to horizontal w/2, and a completely remote sensing image is scanned successively；Record what each step scanning was all obtained The position coordinates in the candidate window upper left corner, is used as the positional information of candidate's picture；
S4：Detection sample input detector obtains result；
S4-1：The candidate window that step S3 is obtained trains the input of obtained detector as step S2, to all Candidate window is detected, records the candidate's picture for being judged as including bridge by detector, and preserve these candidate windows；
S4-2：The positional information that the candidate window of preservation is included is extracted, then the basis on picture to be detected The positional information of candidate window marks the image-region representated by candidate window, is finally completed to remote sensing images Bridge position Detection work.
The step S1-1 should choose the obvious picture of bridge feature when bridge picture is intercepted, while also Interception includes bridge, but feature is not obvious, is blocked or more fuzzy bridge picture.
The convolutional neural networks model that the step S2-1 is set up includes input layer, convolutional layer, pond layer, convolutional layer, pond Change layer, full articulamentum and output layer；
1) input layers are, as input, to be input to positive negative sample in convolutional neural networks model；
2) the feature extractions first stage：The convolution kernel size of convolutional layer is 5*5, inputs 3 passages, exports 64 passages, is moved Dynamic step-length is 1；Pond layer is carried out by the way of maximum pond, and window size is 3*3, and step-length is 2, then by obtained feature Figure is normalized；
3) enters feature extraction second stage：The convolution kernel size of convolutional layer remains 5*5, inputs 64 passages, output 64 Passage, step-length is 1, then pond will be carried out after the characteristic pattern normalization operation after convolution, pond mode, which remains unchanged, takes maximum Chi Hua, window size is 3*3, and step-length is 2；
4) pond result is finally put into full articulamentum by, is finally exported.
Right value update in the convolutional neural networks model that the step S2-1 is set up is carried out using BP back propagations； The method selection gradient descent method of every layer of renewal weights；The Learning Rate learning rates of the gradient descent method are arranged on Between 0.003-0.004.
The last output for the convolutional neural networks model that the step S2-1 is set up uses Softmax as two graders, Softmax is returned in two steps：The first step is in order to obtain the evidence that a given picture belongs to some optional network specific digit class, to picture picture Plain value is weighted summation；If there is this pixel very strong evidence to illustrate that this pictures is not belonging to such, then corresponding Weights are negative, if this opposite pixel possesses favourable evidence and supports this pictures to belong to this class, then weights are just Number；I.e.：
evidenceiRepresent that given picture belongs to the evidence of i classes；Wherein wiRepresent weight, biRepresent the biasing of numeral i classes Amount, the pixel index that j represents given picture x is summed for pixel；Then these evidences can be converted into Softmax functions Probability y：
Y=softmax (evidence)
Wherein, Softmax is an excitation function, therefore, gives a pictures, it is for each digital goodness of fit One probable value is converted into by Softmax functions；Softmax functions are defined as：
Softmax (x)=normalize (exp (x))
Deploy the minor on the right of equation, obtain：
After the result of a probability distribution is obtained using Softmax graders, result is compared with final label It is right, and a threshold value T is determined by comparing, the threshold value is represented when the probable value in Softmax training results is more than T, then Judge to include bridge in input picture；If the probable value in training result is less than T, that judges not including bridge in input picture Beam.
During repetitive exercise in the step S2-2, the strategy of circuit training is taken；Every time from all samples pictures In randomly select a number of picture and be trained, the batch_size sizes of selection are 128, then randomly select same number Other samples of amount are trained, in constantly cyclic process, gradually update the weights in convolutional neural networks model.
Beneficial effect：Feature extraction of the present invention without carrying out bridge picture in advance, simplifies detecting step, is keeping high The detection speed of remote sensing images is greatly accelerated while verification and measurement ratio.
Brief description of the drawings
Fig. 1 is the inventive method flow chart.
Fig. 2 is convolutional neural networks structure chart.
Embodiment
The present invention is further described below in conjunction with the accompanying drawings.
The present invention is a kind of remote sensing images Bridges Detection based on convolutional neural networks, mainly include the training stage and Detection-phase, the described training stage mainly includes the following steps that：
S1:Training sample is gathered and pretreatment；
S2:Convolutional neural networks training pattern is set up, detector is obtained.
Described detection-phase is mainly included the following steps that：
S3:Detect the pretreatment of sample；
S4:Detection sample input detector obtains result.
Further, described step S1 includes following sub-step：
S1-1:A part of remote sensing images are chosen first, and manual interception size size is the bridge of w*h sizes on remote sensing images Beam picture, it is necessary to choose the obvious picture of bridge feature when bridge picture is intercepted, meanwhile, it should also intercept some bags Containing bridge, but feature is not obvious, is blocked or more fuzzy bridge picture, can so ensure in training positive sample Afterwards, detector also has certain detectability to the unconspicuous bridge picture of feature.
S1-2:The region of bridge is not included on remote sensing images, also interception size size is w*h picture, these pictures It is trained as the negative sample of detector.
S1-3：The positive negative sample obtained in selecting step S1-1, S1-2, on the premise of picture w*h sizes are kept, Align negative sample picture and carry out flip horizontal, change of scale, translation transformation, rotation transformation and whitening operation, so do further The quantity of training sample is added, while also allowing the feature of training picture to become more.
Further, described step S2 includes following sub-step：
S2-1；Convolutional neural networks model should be set up first, and the structure of whole model is input layer, convolutional layer, Chi Hua Layer, normalizing layer, convolutional layer normalizes layer, pond layer, full articulamentum, output layer.
1) input layers are, as input, to be input to positive negative sample in convolutional neural networks model；
2) the feature extractions first stage, the convolution kernel size of convolutional layer is 5*5, inputs 3 passages, exports 64 passages, is moved Dynamic step-length is 1, and pond layer is carried out by the way of maximum pond, and window size is 3*3, and step-length is 2, then by obtained feature Figure is normalized；
3) enters feature extraction second stage, and the convolution kernel size of convolutional layer remains 5*5, inputs 64 passages, output 64 Passage, step-length is 1, then pond will be carried out after the characteristic pattern normalization operation after convolution, pond mode, which remains unchanged, takes maximum Chi Hua, window size is 3*3, and step-length is 2；
4) pond result is finally put into full articulamentum by, is finally exported.As shown in Figure 2.
S2-2:Need to carry out just the parameters in network model after the model structure of convolutional neural networks is designed Beginningization, in data initialization, randomness is as high as possible, and convergent speed can be than very fast when so training, and is not easy to be absorbed in The result of local optimum.
S2-3:Right value update in convolutional neural networks model is carried out using BP back propagations, BP back propagation roots The result calculated according to propagated forward is mutually compared with objective result, the difference between two results, i.e. overall error is drawn, according to total Error progressively forward, updates each layer of weights.The method selection gradient descent method of weights is updated at every layer, is declined using gradient Method calculates the best initial weights under error current, obtains updating the weights of front layer successively after best initial weights.Gradient descent method Learning Rate learning rates are arranged between 0.003-0.004.
S2-4:Finally output is using Softmax as two graders, and Softmax is returned in two steps：The first step is in order to obtain One given picture belongs to the evidence (evidence) of some optional network specific digit class, and we are weighted summation to picture pixels value. If there is this pixel very strong evidence to illustrate that this pictures is not belonging to such, then corresponding weights are negative, conversely such as Really this pixel possesses favourable evidence and supports this pictures to belong to this class, then weights are positive numbers.I.e.:
Wherein wiRepresent weight, biThe amount of bias of numeral i classes is represented, the pixel index that j represents given picture x is used for pixel Summation.Then these evidences can be converted into probability y with Softmax functions：
Y=softmax (evidence)
Here Softmax can regard excitation (activation) function as, therefore, give a pictures, it A probable value can be converted into by Softmax functions for each digital goodness of fit.Softmax functions can be defined For：
Softmax (x)=normalize (exp (x))
Deploy the minor on the right of equation, can obtain：
Softmax graders are that the output valve using full articulamentum treats as power exponent evaluation as input value, and input value, These end values of regularization again.This power operation represents that bigger evidence corresponds to the multiplier weight inside bigger hypothesized model Value.
Conversely, possessing less evidence means to possess smaller multiplier coefficients inside hypothesized model.In hypothesized model Weights cannot be 0 value or negative value.Softmax then can regularization these weighted values, their summation is equal to 1, with This one effective probability distribution of construction.
S2-5:After the result of a probability distribution is obtained using Softmax graders, by these results and final mark Label are compared, and determine a threshold value T by comparing, and the probable value that the threshold value represents to work as in Softmax training results is more than T When, then it will judge in input picture it is to include bridge, if the probable value in training result is less than T, that judges input It is not include bridge in picture.
S2-6:Using step S1-1, the positive negative sample that S1-2 is drawn is rolled up after these samples are put into each layer parameter of initialization Product neural network model, is iterated training.In the training process, it will not take and disposably all put all training samples Enter the strategy in model, can so make mode input excessive, calculating the slow and general equipment of meeting can not support yet.
Therefore, the strategy of circuit training can be taken in training process, is randomly selected every time from all samples pictures certain The picture of quantity is trained, and batch size batch_size sizes chosen here are 128, are then randomly selected same amount of Other samples are trained, and in constantly cyclic process (cycle-index is set to 10000), gradually update convolutional neural networks Weights in model, training speed and efficiency can not only be improved by so doing, while accuracy rate also can be higher.
Further, described step S3 includes following sub-step：
S3-1:Remote sensing image to be detected is chosen first, and for general remote sensing images, the size of remote sensing images is all non- Chang great, therefore 1/8th or 1/10th of a remote sensing image to be detected can be intercepted in the present invention, it is detected every time In a part then after detect other parts again, burden of such detector in detection can be smaller, it is easier to counts Calculate, efficiency also can be higher.
S3-2:Remote sensing image to be detected is chosen, is scanned by w*h size windows since the upper left corner of remote sensing image, Transversal scanning step-length is w/2, and when the low order end of scanning to picture to be detected, one is moved down according to longitudinal scanning step-length h/2 OK, then the step scan since Far Left according to horizontal w/2, a completely remote sensing image is scanned successively.
S3-3:Each step scanning all obtains a candidate window, in scanning, by the upper left corner of each candidate window Position coordinates is recorded, and is used as the positional information of candidate's picture.Therefore, each candidate window should be candidate's window comprising information The picture region image that mouth is represented, top left co-ordinate (x, y), wide height (w, h), i.e. (image, x, the y, w, h) of candidate's picture.
Further, described step S4 includes following sub-step：
S4-1:The candidate window that step S3 is obtained trains the input of obtained detector as step S2, to all Candidate window is detected, records the candidate's picture for being judged as including bridge by detector, and preserve these candidate windows.
S4-2:The positional information that the candidate window of preservation is included is extracted, then the basis on picture to be detected The positional information of candidate window marks the image-region representated by candidate window, is finally completed to remote sensing images Bridge position Detection work.
Due to the use of CPU calculating speeds being slow for GPU, therefore GPU has been used to be trained finally And calculating, this causes training speed to be greatly improved, while detection efficiency is also greatly improved.
Described above is only the preferred embodiment of the present invention, it should be pointed out that：For the ordinary skill people of the art For member, under the premise without departing from the principles of the invention, some improvements and modifications can also be made, these improvements and modifications also should It is considered as protection scope of the present invention.

Claims (6)

1. a kind of remote sensing images Bridges Detection based on convolutional neural networks, it is characterised in that：Including step：
S1：Training sample is gathered and pretreatment；
S1-1：The remote sensing images for including bridge area are chosen, manual interception size size is the bridge of w*h sizes on remote sensing images Beam picture；
S1-2：The region of bridge is not included on remote sensing images, interception size size is w*h picture, is used as the negative of detector Sample is trained；
S1-3：The positive negative sample obtained in selecting step S1-1, S1-2, on the premise of picture w*h sizes are kept, is aligned Negative sample picture carries out flip horizontal, change of scale, translation transformation, rotation transformation and whitening operation；
S2：Convolutional neural networks training pattern is set up, detector is obtained；
S2-1：Convolutional neural networks model is set up, and the parameters in convolutional neural networks model are initialized；
S2-2：The positive negative sample that step S1-1, S1-2 is obtained is put into the convolutional neural networks model that S2-1 is obtained, and is iterated Training；
S3：Detect the pretreatment of sample：
Remote sensing image to be detected is chosen, is scanned by w*h size windows since the upper left corner of remote sensing image, transversal scanning step A length of w/2, when the low order end of scanning to picture to be detected, a line is moved down according to longitudinal scanning step-length h/2, then from most left While starting the step scan according to horizontal w/2, a completely remote sensing image is scanned successively；Record the candidate that each step scanning is all obtained The position coordinates in the window upper left corner, is used as the positional information of candidate's picture；
S4：Detection sample input detector obtains result；
S4-1：The candidate window that step S3 is obtained trains the input of obtained detector as step S2, to all candidates Window is detected, records the candidate's picture for being judged as including bridge by detector, and preserve these candidate windows；
S4-2：The positional information that the candidate window of preservation is included is extracted, then according to candidate on picture to be detected The positional information of window marks the image-region representated by candidate window, is finally completed the inspection to remote sensing images Bridge position Survey work.
2. remote sensing images Bridges Detection according to claim 1, it is characterised in that：The step S1-1 is in interception bridge When beam picture, the obvious picture of bridge feature should be chosen, while also to intercept comprising bridge, but feature is not obvious, It is blocked or more fuzzy bridge picture.
3. remote sensing images Bridges Detection according to claim 1, it is characterised in that：The volume that the step S2-1 is set up Product neural network model includes input layer, convolutional layer, pond layer, convolutional layer, pond layer, full articulamentum and output layer；
1) input layers are, as input, to be input to positive negative sample in convolutional neural networks model；
2) the feature extractions first stage：The convolution kernel size of convolutional layer is 5*5, inputs 3 passages, exports 64 passages, mobile step A length of 1；Pond layer is carried out by the way of maximum pond, and window size is 3*3, and step-length is 2, then enters obtained characteristic pattern Row normalization；
3) enters feature extraction second stage：The convolution kernel size of convolutional layer remains 5*5, inputs 64 passages, and output 64 is led to Road, step-length is 1, then pond will be carried out after the characteristic pattern normalization operation after convolution, pond mode, which remains unchanged, takes maximum pond Change, window size is 3*3, and step-length is 2；
4) pond result is finally put into full articulamentum by, is finally exported.
4. remote sensing images Bridges Detection according to claim 1, it is characterised in that：The volume that the step S2-1 is set up Right value update in product neural network model is carried out using BP back propagations；Under every layer of method selection gradient for updating weights Drop method；The Learning Rate learning rates of the gradient descent method are arranged between 0.003-0.004.
5. remote sensing images Bridges Detection according to claim 1, it is characterised in that：The volume that the step S2-1 is set up The last output of product neural network model is using Softmax as two graders, and Softmax is returned in two steps：The first step in order to The evidence that a given picture belongs to some optional network specific digit class is obtained, summation is weighted to picture pixels value；If this picture There is element very strong evidence to illustrate that this pictures is not belonging to such, then corresponding weights are negative, if this opposite pixel Possessing favourable evidence supports this pictures to belong to this class, then weights are positive numbers；I.e.：
<mrow> <msub> <mi>evidence</mi> <mi>i</mi> </msub> <mo>=</mo> <munder> <mi>&amp;Sigma;</mi> <mi>j</mi> </munder> <msub> <mi>w</mi> <mrow> <mi>i</mi> <mo>,</mo> <mi>j</mi> </mrow> </msub> <msub> <mi>x</mi> <mi>j</mi> </msub> <mo>+</mo> <msub> <mi>b</mi> <mi>i</mi> </msub> </mrow>
evidenceiRepresent that given picture belongs to the evidence of i classes；Wherein wiRepresent weight, biRepresent the amount of bias of numeral i classes, j The pixel index for representing given picture x is summed for pixel；Then these evidences can be converted into probability with Softmax functions y：
Y=softmax (evidence)
Wherein, Softmax is an excitation function, therefore, gives a pictures, it is for each digital goodness of fit quilt Softmax functions are converted into a probable value；Softmax functions are defined as：
Softmax (x)=normalize (exp (x))
Deploy the minor on the right of equation, obtain：
<mrow> <mi>s</mi> <mi>o</mi> <mi>f</mi> <mi>t</mi> <mi>max</mi> <msub> <mrow> <mo>(</mo> <mi>x</mi> <mo>)</mo> </mrow> <mi>i</mi> </msub> <mo>=</mo> <mfrac> <mrow> <mi>exp</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>i</mi> </msub> <mo>)</mo> </mrow> </mrow> <mrow> <msub> <mi>&amp;Sigma;</mi> <mi>j</mi> </msub> <mi>exp</mi> <mrow> <mo>(</mo> <msub> <mi>x</mi> <mi>j</mi> </msub> <mo>)</mo> </mrow> </mrow> </mfrac> <mo>;</mo> </mrow>
After the result of a probability distribution is obtained using Softmax graders, result is compared with final label, and A threshold value T is determined by comparing, the threshold value is represented when the probable value in Softmax training results is more than T, then judged defeated Enter and bridge is included in picture；If the probable value in training result is less than T, that judges not including bridge in input picture.
6. remote sensing images Bridges Detection according to claim 1, it is characterised in that：Iteration in the step S2-2 In training process, the strategy of circuit training is taken；A number of picture is randomly selected from all samples pictures every time to carry out Training, the batch_size sizes of selection are 128, then randomly select other same amount of samples and are trained, continuous In ground cyclic process, the weights in convolutional neural networks model are gradually updated.
CN201710380211.6A 2017-05-25 2017-05-25 A kind of remote sensing images Bridges Detection based on convolutional neural networks Pending CN107066995A (en)

Priority Applications (2)

Application Number Priority Date Filing Date Title
CN201710380211.6A CN107066995A (en) 2017-05-25 2017-05-25 A kind of remote sensing images Bridges Detection based on convolutional neural networks
PCT/CN2017/089134 WO2018214195A1 (en) 2017-05-25 2017-06-20 Remote sensing imaging bridge detection method based on convolutional neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710380211.6A CN107066995A (en) 2017-05-25 2017-05-25 A kind of remote sensing images Bridges Detection based on convolutional neural networks

Publications (1)

Publication Number Publication Date
CN107066995A true CN107066995A (en) 2017-08-18

Family

ID=59609850

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710380211.6A Pending CN107066995A (en) 2017-05-25 2017-05-25 A kind of remote sensing images Bridges Detection based on convolutional neural networks

Country Status (2)

Country Link
CN (1) CN107066995A (en)
WO (1) WO2018214195A1 (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108364262A (en) * 2018-01-11 2018-08-03 深圳大学 A kind of restored method of blurred picture, device, equipment and storage medium
CN109086878A (en) * 2018-10-19 2018-12-25 电子科技大学 Keep the convolutional neural networks model and its training method of rotational invariance
CN109325449A (en) * 2018-01-04 2019-02-12 苏州中科天启遥感科技有限公司 Convolutional neural networks target detection frame based on Sample Refreshment
CN109961083A (en) * 2017-12-14 2019-07-02 安讯士有限公司 For convolutional neural networks to be applied to the method and image procossing entity of image
CN110675324A (en) * 2018-07-02 2020-01-10 上海寰声智能科技有限公司 4K ultra-high definition image sharpening processing method
WO2020147345A1 (en) * 2019-01-14 2020-07-23 珠海格力电器股份有限公司 Method and device for obtaining chalkiness of rice grain and cooking appliance
CN111815627A (en) * 2020-08-24 2020-10-23 成都睿沿科技有限公司 Remote sensing image change detection method, model training method and corresponding device

Families Citing this family (107)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111368865B (en) * 2018-12-26 2023-08-22 北京眼神智能科技有限公司 Remote sensing image oil storage tank detection method and device, readable storage medium and equipment
CN111382759B (en) * 2018-12-28 2023-04-21 广州市百果园信息技术有限公司 Pixel classification method, device, equipment and storage medium
CN109740549B (en) * 2019-01-08 2022-12-27 西安电子科技大学 SAR image target detection system and method based on semi-supervised CNN
CN111476056B (en) * 2019-01-23 2024-04-16 阿里巴巴集团控股有限公司 Target object identification method, device, terminal equipment and computer storage medium
CN109859187B (en) * 2019-01-31 2023-04-07 东北大学 Explosive-pile ore rock particle image segmentation method
CN109919230B (en) * 2019-03-10 2022-12-06 西安电子科技大学 Medical image pulmonary nodule detection method based on cyclic feature pyramid
CN109919108B (en) * 2019-03-11 2022-12-06 西安电子科技大学 Remote sensing image rapid target detection method based on deep hash auxiliary network
CN109978032B (en) * 2019-03-15 2022-12-06 西安电子科技大学 Bridge crack detection method based on space pyramid cavity convolution network
CN110009010B (en) * 2019-03-20 2023-03-24 西安电子科技大学 Wide-width optical remote sensing target detection method based on interest area redetection
CN110147714B (en) * 2019-03-28 2023-06-23 煤炭科学研究总院 Unmanned aerial vehicle-based coal mine goaf crack identification method and detection system
CN109993104B (en) * 2019-03-29 2022-09-16 河南工程学院 Method for detecting change of object level of remote sensing image
CN110020688B (en) * 2019-04-10 2022-12-06 西安电子科技大学 Shielded pedestrian detection method based on deep learning
CN110334724B (en) * 2019-04-16 2022-06-17 武汉理工大学 Remote sensing object natural language description and multi-scale correction method based on LSTM
CN110097524B (en) * 2019-04-22 2022-12-06 西安电子科技大学 SAR image target detection method based on fusion convolutional neural network
CN110210297B (en) * 2019-04-25 2023-12-26 上海海事大学 Method for locating and extracting Chinese characters in customs clearance image
CN110084195B (en) * 2019-04-26 2022-12-06 西安电子科技大学 Remote sensing image target detection method based on convolutional neural network
CN110334578B (en) * 2019-05-05 2023-04-18 中南大学 Weak supervision method for automatically extracting high-resolution remote sensing image buildings through image level annotation
CN110288563A (en) * 2019-05-22 2019-09-27 苏州万卓纺织有限公司 A kind of fabric defect detection method based on deep learning
CN110321794B (en) * 2019-05-23 2023-02-28 湖南大学 Remote sensing image oil tank detection method integrated with semantic model
CN111985274B (en) * 2019-05-23 2023-08-04 中国科学院沈阳自动化研究所 Remote sensing image segmentation method based on convolutional neural network
CN110163294B (en) * 2019-05-29 2023-05-09 广东工业大学 Remote sensing image change region detection method based on dimension reduction operation and convolution network
CN110414330B (en) * 2019-06-20 2023-05-26 平安科技（深圳）有限公司 Palm image detection method and device
TWI738009B (en) * 2019-06-20 2021-09-01 和碩聯合科技股份有限公司 Object detection system and object detection method
CN110334765B (en) * 2019-07-05 2023-03-24 西安电子科技大学 Remote sensing image classification method based on attention mechanism multi-scale deep learning
CN110321866B (en) * 2019-07-09 2023-03-24 西北工业大学 Remote sensing image scene classification method based on depth feature sparsification algorithm
CN110472636B (en) * 2019-07-26 2022-10-14 四创科技有限公司 Deep learning-based water gauge E-shaped scale identification method
CN110443259B (en) * 2019-07-29 2023-04-07 中科光启空间信息技术有限公司 Method for extracting sugarcane from medium-resolution remote sensing image
CN112347827A (en) * 2019-08-06 2021-02-09 东北大学秦皇岛分校 Automatic detection method and system for ship water gauge
CN112395924B (en) * 2019-08-16 2024-02-20 阿里巴巴集团控股有限公司 Remote sensing monitoring method and device
CN110472732B (en) * 2019-08-19 2023-02-21 杭州凝眸智能科技有限公司 Image feature extraction system based on optimized feature extraction device
CN110569899B (en) * 2019-09-03 2022-06-10 嘉陵江亭子口水利水电开发有限公司 Dam face defect classification model training method and device
CN112446266B (en) * 2019-09-04 2024-03-29 北京君正集成电路股份有限公司 Face recognition network structure suitable for front end
CN110648334A (en) * 2019-09-18 2020-01-03 中国人民解放***箭军工程大学 Multi-feature cyclic convolution saliency target detection method based on attention mechanism
CN110728665B (en) * 2019-09-30 2023-04-18 西安电子科技大学 SAR image change detection method based on parallel probabilistic neural network
CN110751271B (en) * 2019-10-28 2023-05-26 西安烽火软件科技有限公司 Image traceability feature characterization method based on deep neural network
CN111047551B (en) * 2019-11-06 2023-10-31 北京科技大学 Remote sensing image change detection method and system based on U-net improved algorithm
CN111008956A (en) * 2019-11-13 2020-04-14 武汉工程大学 Beam bottom crack detection method, system, device and medium based on image processing
CN111008651B (en) * 2019-11-13 2023-04-28 科大国创软件股份有限公司 Image reproduction detection method based on multi-feature fusion
CN110991374B (en) * 2019-12-10 2023-04-04 电子科技大学 Fingerprint singular point detection method based on RCNN
CN111126189A (en) * 2019-12-10 2020-05-08 北京航天世景信息技术有限公司 Target searching method based on remote sensing image
CN111104887B (en) * 2019-12-11 2024-03-29 北京化工大学 Full-period keyless phase monitoring method based on vibration mechanism and deep learning technology
CN111160127B (en) * 2019-12-11 2023-07-21 中国四维测绘技术有限公司 Remote sensing image processing and detecting method based on deep convolutional neural network model
CN111028178B (en) * 2019-12-20 2022-04-29 武汉大学 Remote sensing image data automatic geometric correction method based on deep learning
CN110992257A (en) * 2019-12-20 2020-04-10 北京航天泰坦科技股份有限公司 Remote sensing image sensitive information automatic shielding method and device based on deep learning
CN111192240B (en) * 2019-12-23 2023-09-01 北京航空航天大学 Remote sensing image target detection method based on random access memory
CN110988839B (en) * 2019-12-25 2023-10-10 中南大学 Quick identification method for wall health condition based on one-dimensional convolutional neural network
CN111241725B (en) * 2019-12-30 2022-08-23 浙江大学 Structure response reconstruction method for generating countermeasure network based on conditions
CN111160276B (en) * 2019-12-31 2023-05-12 重庆大学 U-shaped cavity full convolution segmentation network identification model based on remote sensing image
CN111199214B (en) * 2020-01-04 2023-05-05 西安电子科技大学 Residual network multispectral image ground object classification method
CN111222576B (en) * 2020-01-08 2023-03-24 西安理工大学 High-resolution remote sensing image classification method
CN111144383B (en) * 2020-01-15 2023-03-28 河南理工大学 Method for detecting vehicle deflection angle
CN111310675A (en) * 2020-02-20 2020-06-19 上海赛可出行科技服务有限公司 Overhead identification auxiliary positioning method based on convolutional neural network
CN111339935B (en) * 2020-02-25 2023-04-18 西安电子科技大学 Optical remote sensing picture classification method based on interpretable CNN image classification model
CN111353432B (en) * 2020-02-28 2023-08-01 安徽华润金蟾药业股份有限公司 Rapid clean selection method and system for honeysuckle medicinal materials based on convolutional neural network
CN111401190A (en) * 2020-03-10 2020-07-10 上海眼控科技股份有限公司 Vehicle detection method, device, computer equipment and storage medium
CN111368776B (en) * 2020-03-13 2024-03-22 长安大学 High-resolution remote sensing image classification method based on deep ensemble learning
CN111401302B (en) * 2020-04-07 2022-08-02 中国人民解放军海军航空大学 Remote sensing image ship target integrated detection and fine-grained identification method
CN111489387B (en) * 2020-04-09 2023-10-20 湖南盛鼎科技发展有限责任公司 Remote sensing image building area calculation method
CN111476167B (en) * 2020-04-09 2024-03-22 北京中科千寻科技有限公司 One-stage direction remote sensing image target detection method based on student-T distribution assistance
CN111553303B (en) * 2020-05-07 2024-03-29 武汉大势智慧科技有限公司 Remote sensing orthographic image dense building extraction method based on convolutional neural network
CN111626175B (en) * 2020-05-22 2023-05-19 西安工业大学 Shaft type identification method based on deep convolutional neural network
CN111754463B (en) * 2020-06-02 2024-05-14 石家庄铁道大学 Method for detecting CA mortar layer defects of ballastless track based on convolutional neural network
CN111666903B (en) * 2020-06-10 2022-10-04 中国电子科技集团公司第二十八研究所 Method for identifying thunderstorm cloud cluster in satellite cloud picture
CN111709479B (en) * 2020-06-17 2022-05-10 广东工业大学 Image classification method and device
CN111950343B (en) * 2020-06-24 2024-03-22 中国电力科学研究院有限公司 Automatic transmission tower identification method and system based on satellite remote sensing technology
CN111914997B (en) * 2020-06-30 2024-04-02 华为技术有限公司 Method for training neural network, image processing method and device
CN111986149A (en) * 2020-07-16 2020-11-24 江西斯源科技有限公司 Plant disease and insect pest detection method based on convolutional neural network
CN111915592B (en) * 2020-08-04 2023-08-22 西安电子科技大学 Remote sensing image cloud detection method based on deep learning
CN111985549B (en) * 2020-08-12 2023-03-31 中国科学院光电技术研究所 Deep learning method for automatic positioning and identification of components for given rigid body target
CN112084877B (en) * 2020-08-13 2023-08-18 西安理工大学 NSGA-NET-based remote sensing image recognition method
CN112132759B (en) * 2020-09-07 2024-03-19 东南大学 Skull face restoration method based on end-to-end convolutional neural network
CN112053354B (en) * 2020-09-15 2024-01-30 上海应用技术大学 Rail plate crack detection method
CN112132822B (en) * 2020-09-30 2024-05-07 东南大学 Suspicious illegal building detection algorithm based on transfer learning
CN112232229B (en) * 2020-10-20 2022-04-01 山东科技大学 Fine water body extraction method based on U-net neural network
CN112751633B (en) * 2020-10-26 2022-08-26 中国人民解放军63891部队 Broadband spectrum detection method based on multi-scale window sliding
CN112651931B (en) * 2020-12-15 2024-04-26 浙江大华技术股份有限公司 Building deformation monitoring method and device and computer equipment
CN112801972A (en) * 2021-01-25 2021-05-14 武汉理工大学 Bridge defect detection method, device, system and storage medium
CN112784806A (en) * 2021-02-04 2021-05-11 中国地质科学院矿产资源研究所 Lithium-containing pegmatite vein extraction method based on full convolution neural network
CN112906577B (en) * 2021-02-23 2024-04-26 清华大学 Fusion method of multisource remote sensing images
CN113033324B (en) * 2021-03-03 2024-03-08 广东省地质环境监测总站 Geological disaster precursor factor identification method and device, electronic equipment and storage medium
CN113095359B (en) * 2021-03-05 2023-09-12 西安交通大学 Method and system for detecting radiographic image marking information
CN113160239B (en) * 2021-03-08 2023-09-22 广东国地规划科技股份有限公司 Illegal land detection method and device
CN113111740A (en) * 2021-03-27 2021-07-13 西北工业大学 Characteristic weaving method for remote sensing image target detection
CN113505627A (en) * 2021-03-31 2021-10-15 北京苍灵科技有限公司 Remote sensing data processing method and device, electronic equipment and storage medium
CN113255451B (en) * 2021-04-25 2023-04-07 西北工业大学 Method and device for detecting change of remote sensing image, electronic equipment and storage medium
CN113344037A (en) * 2021-05-18 2021-09-03 国网江西省电力有限公司电力科学研究院 Cable insulation layer and sheath parameter measuring method and measuring device
CN113408398B (en) * 2021-06-16 2023-04-07 西安电子科技大学 Remote sensing image cloud detection method based on channel attention and probability up-sampling
CN113536943B (en) * 2021-06-21 2024-04-12 上海赫千电子科技有限公司 Road traffic sign recognition method based on image enhancement
CN113570111B (en) * 2021-06-29 2023-08-29 中北大学 Bridge health state on-chip monitoring method based on lightweight network
CN113538615B (en) * 2021-06-29 2024-01-09 中国海洋大学 Remote sensing image coloring method based on double-flow generator depth convolution countermeasure generation network
CN113537006A (en) * 2021-07-01 2021-10-22 昆明理工大学 Pu-erh raw tea and ripe tea judging method based on convolutional neural network
CN113469072B (en) * 2021-07-06 2024-04-12 西安电子科技大学 Remote sensing image change detection method and system based on GSoP and twin fusion network
CN113343942B (en) * 2021-07-21 2023-05-23 西安电子科技大学 Remote sensing image defect detection method
CN113361662B (en) * 2021-07-22 2023-08-29 全图通位置网络有限公司 Urban rail transit remote sensing image data processing system and method
CN113837994B (en) * 2021-07-29 2024-02-13 尚特杰电力科技有限公司 Photovoltaic panel defect diagnosis method based on edge detection convolutional neural network
CN113706364A (en) * 2021-09-14 2021-11-26 杭州师范大学 Reversible information hiding method for remote sensing image
CN113792685B (en) * 2021-09-17 2024-03-12 东北石油大学 Microseism event detection method based on multi-scale convolutional neural network
CN114092832B (en) * 2022-01-20 2022-04-15 武汉大学 High-resolution remote sensing image classification method based on parallel hybrid convolutional network
CN114519819B (en) * 2022-02-10 2024-04-02 西北工业大学 Remote sensing image target detection method based on global context awareness
CN114821056A (en) * 2022-04-24 2022-07-29 国家林业和草原局华东调查规划院 AI technology-based forest and grass resource change automatic interpretation method in remote sensing image
CN114782400B (en) * 2022-05-17 2023-06-20 东风本田发动机有限公司 Method, device, equipment, medium and program product for detecting slag point of metal material
CN114657513B (en) * 2022-05-23 2022-09-20 河南银金达新材料股份有限公司 Preparation method of antibacterial regenerated polyester film
CN115096268B (en) * 2022-06-17 2023-06-30 西南交通大学 Bridge damage depth detection method based on unmanned aerial vehicle aerial photography and ultrasonic detection
CN115423829B (en) * 2022-07-29 2024-03-01 江苏省水利科学研究院 Method and system for rapidly extracting water body of single-band remote sensing image
CN114998772B (en) * 2022-08-03 2022-11-18 深圳联和智慧科技有限公司 Integrated bridge detection method and system based on unmanned aerial vehicle and cloud platform
CN116645618B (en) * 2023-06-05 2023-12-08 广东省农业科学院设施农业研究所 Agricultural data processing method, system and storage medium
CN117036962B (en) * 2023-10-08 2024-02-06 中国科学院空天信息创新研究院 Remote sensing image change detection method, device, equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106096655A (en) * 2016-06-14 2016-11-09 厦门大学 A kind of remote sensing image airplane detection method based on convolutional neural networks
CN106295503A (en) * 2016-07-25 2017-01-04 武汉大学 The high-resolution remote sensing image Ship Target extracting method of region convolutional neural networks
JP2017062776A (en) * 2015-09-04 2017-03-30 株式会社東芝 Method and device for detecting changes in structure, and computer readable medium

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9202144B2 (en) * 2013-10-30 2015-12-01 Nec Laboratories America, Inc. Regionlets with shift invariant neural patterns for object detection
CN104217214B (en) * 2014-08-21 2017-09-19 广东顺德中山大学卡内基梅隆大学国际联合研究院 RGB D personage's Activity recognition methods based on configurable convolutional neural networks
CN105844228B (en) * 2016-03-21 2019-02-19 北京航空航天大学 A kind of remote sensing images cloud detection method of optic based on convolutional neural networks

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017062776A (en) * 2015-09-04 2017-03-30 株式会社東芝 Method and device for detecting changes in structure, and computer readable medium
CN106096655A (en) * 2016-06-14 2016-11-09 厦门大学 A kind of remote sensing image airplane detection method based on convolutional neural networks
CN106295503A (en) * 2016-07-25 2017-01-04 武汉大学 The high-resolution remote sensing image Ship Target extracting method of region convolutional neural networks

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
搬砖小工053: "[03]tensorflow实现softmax回归（softmax regression）", 《HTTPS://BLOG.CSDN.NET/SA14023053/ARTICLE/DETAILS/51884894》 *
欧阳颖卉 等: "基于卷积神经网络的光学遥感图像船只检测", 《包装工程》 *

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109961083A (en) * 2017-12-14 2019-07-02 安讯士有限公司 For convolutional neural networks to be applied to the method and image procossing entity of image
CN109325449A (en) * 2018-01-04 2019-02-12 苏州中科天启遥感科技有限公司 Convolutional neural networks target detection frame based on Sample Refreshment
CN108364262A (en) * 2018-01-11 2018-08-03 深圳大学 A kind of restored method of blurred picture, device, equipment and storage medium
CN110675324A (en) * 2018-07-02 2020-01-10 上海寰声智能科技有限公司 4K ultra-high definition image sharpening processing method
CN110675324B (en) * 2018-07-02 2023-10-10 上海寰声智能科技有限公司 4K ultra-high definition image sharpening processing method
CN109086878A (en) * 2018-10-19 2018-12-25 电子科技大学 Keep the convolutional neural networks model and its training method of rotational invariance
CN109086878B (en) * 2018-10-19 2019-12-17 电子科技大学 convolutional neural network model keeping rotation invariance and training method thereof
WO2020147345A1 (en) * 2019-01-14 2020-07-23 珠海格力电器股份有限公司 Method and device for obtaining chalkiness of rice grain and cooking appliance
CN111815627A (en) * 2020-08-24 2020-10-23 成都睿沿科技有限公司 Remote sensing image change detection method, model training method and corresponding device

Also Published As

Publication number Publication date
WO2018214195A1 (en) 2018-11-29

Similar Documents

Publication Publication Date Title
CN107066995A (en) A kind of remote sensing images Bridges Detection based on convolutional neural networks
CN110084195B (en) Remote sensing image target detection method based on convolutional neural network
CN109409263B (en) Method for detecting urban ground feature change of remote sensing image based on Siamese convolutional network
CN102842045B (en) A kind of pedestrian detection method based on assemblage characteristic
CN108596101A (en) A kind of remote sensing images multi-target detection method based on convolutional neural networks
CN107742099A (en) A kind of crowd density estimation based on full convolutional network, the method for demographics
CN109614985A (en) A kind of object detection method based on intensive connection features pyramid network
CN108830188A (en) Vehicle checking method based on deep learning
CN109711288A (en) Remote sensing ship detecting method based on feature pyramid and distance restraint FCN
CN108256424A (en) A kind of high-resolution remote sensing image method for extracting roads based on deep learning
CN107123123A (en) Image segmentation quality evaluating method based on convolutional neural networks
CN109934115A (en) Construction method, face identification method and the electronic equipment of human face recognition model
CN106934795A (en) The automatic testing method and Forecasting Methodology of a kind of glue into concrete beam cracks
CN109766936A (en) Image change detection method based on information transmitting and attention mechanism
CN106778683A (en) Based on the quick Multi-angle face detection method for improving LBP features
Bou Kheir et al. The application of GIS based decision-tree models for generating the spatial distribution of hydromorphic organic landscapes in relation to digital terrain data
CN112861732B (en) Method, system and device for monitoring land in ecological environment fragile area
CN109635726A (en) A kind of landslide identification method based on the symmetrical multiple dimensioned pond of depth network integration
CN106228130A (en) Remote sensing image cloud detection method of optic based on fuzzy autoencoder network
CN105654477B (en) A kind of detecting and positioning method of ribbon buried target
CN107895136A (en) A kind of colliery area recognizing method and system
CN115512247A (en) Regional building damage grade assessment method based on image multi-parameter extraction
CN109785359A (en) A kind of video object detection method based on depth characteristic pyramid and tracking loss
CN112084860A (en) Target object detection method and device and thermal power plant detection method and device
CN113469097B (en) Multi-camera real-time detection method for water surface floaters based on SSD network

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication
RJ01 Rejection of invention patent application after publication

Application publication date: 20170818