CN115937169A - Shrimp fry counting method and system based on high resolution and target detection - Google Patents
Shrimp fry counting method and system based on high resolution and target detection Download PDFInfo
- Publication number
- CN115937169A CN115937169A CN202211666456.2A CN202211666456A CN115937169A CN 115937169 A CN115937169 A CN 115937169A CN 202211666456 A CN202211666456 A CN 202211666456A CN 115937169 A CN115937169 A CN 115937169A
- Authority
- CN
- China
- Prior art keywords
- shrimp
- detection
- larvae
- counting
- resolution
- 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
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 123
- 238000000034 method Methods 0.000 title claims abstract description 42
- 241000143060 Americamysis bahia Species 0.000 title 1
- 241000238557 Decapoda Species 0.000 claims abstract description 226
- 238000012549 training Methods 0.000 claims abstract description 40
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 16
- 238000005520 cutting process Methods 0.000 claims description 10
- 238000002372 labelling Methods 0.000 claims description 8
- 241000238552 Penaeus monodon Species 0.000 abstract description 4
- 230000000694 effects Effects 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000011160 research Methods 0.000 description 2
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 241000530454 Litopenaeus schmitti Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
- Y02A40/80—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in fisheries management
- Y02A40/81—Aquaculture, e.g. of fish
Landscapes
- Image Analysis (AREA)
- Farming Of Fish And Shellfish (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a shrimp fry counting method and a shrimp fry counting system based on high resolution and target detection, which comprise a model training stage and an inference use stage, wherein the model training stage comprises the following steps: s11: acquiring shrimp larva images for model training by using a high-resolution industrial camera; s12: carrying out boundary frame marking on the shrimp larva image to obtain a JSON tag file; s13: and inputting the shrimp larva images and the JSON label file into the high-resolution shrimp larva detection model for training to obtain the trained high-resolution shrimp larva detection model. The invention can be suitable for the detection and counting of the shrimp seeds of different varieties including the shrimp seeds of the penaeus monodon, and the accuracy of the detection and counting is improved; in addition, the shrimp larvae image with high resolution enables a large number of shrimp larvae to be counted at one time so as to cope with the situation that the density of the shrimp larvae is high, and therefore counting efficiency is improved.
Description
Technical Field
The invention relates to the technical field of computer vision detection, in particular to a shrimp larvae counting method and system based on high resolution and target detection.
Background
The existing counting methods for the shrimp larvae comprise a traditional counting method based on threshold segmentation and a counting algorithm based on a density map, and the like, wherein the existing algorithm researches take the south America white shrimp larvae as main research objects, and the forms of the shrimp larvae of the variety are relatively consistent; however, when facing the penaeus monodon fries with slender ventral nodes, black and thick outer shells, thick tails and frontal angles of the heads or the fries of other varieties, the detection effect is poor, the counting result has larger errors, and the universality is not high.
In addition, the existing shrimp larva counting method has a good counting effect on low-density shrimp larvae of dozens to hundreds, but has an unsatisfactory counting effect on medium-high-density shrimp larva images of thousands to thousands of shrimps; the more the number of the shrimp seedlings is counted at one time, the denser the shrimp seedlings are, the greater the counting difficulty is, and the lower the accuracy is. Therefore, several seedlings need to be batched a small number of times, which greatly reduces the efficiency.
Disclosure of Invention
Technical problem to be solved
Aiming at the defects of the prior art, the invention provides a shrimp larva counting method and system based on high resolution and target detection, which can solve the technical problems.
(II) technical scheme
In order to solve the above technical problems, the present invention provides the following technical solutions: a shrimp fry counting method based on high resolution and target detection comprises a model training stage and an inference use stage, wherein the model training stage comprises the following steps:
s11: acquiring shrimp larva images for model training by using a high-resolution industrial camera;
s12: carrying out boundary frame marking on the shrimp larva image to obtain a JSON tag file;
s13: inputting the shrimp larva images and the JSON label file into a high-resolution shrimp larva detection model for training to obtain a trained high-resolution shrimp larva detection model;
the inference use phase comprises the following steps:
s21: acquiring a shrimp larva image to be detected by using an industrial camera;
s22: inputting the shrimp larvae image into a trained high-resolution shrimp larvae detection model to perform target detection on the shrimp larvae so as to obtain a shrimp larvae detection result;
s23: and counting the detection result of the shrimp seeds to obtain the counting result of the shrimp seeds.
Preferably, after step S12, the method further includes: and cutting the shrimp larva image to obtain a plurality of sub image blocks.
Preferably, step S13 is specifically: and inputting the plurality of sub image blocks and the JSON label file into a high-resolution shrimp fry detection model for training to obtain a trained high-resolution shrimp fry detection model.
Preferably, before step S22, the method further includes: and cutting the shrimp fry image acquired in the step S21 to obtain a plurality of sub image blocks.
Preferably, step S22 specifically includes the following sub-steps:
s221: inputting a plurality of sub-image blocks into a trained high-resolution shrimp larva detection model;
s222: classifying the plurality of sub image blocks to extract a plurality of shrimp fry detection candidate image blocks;
s223: and carrying out the target detection of the shrimp larvae on the plurality of shrimp larvae detection candidate image blocks to correspondingly obtain the detection results of the plurality of shrimp larvae.
Preferably, step S23 specifically includes: counting and combining the detection results of the plurality of shrimp seeds to obtain the counting result of the shrimp seeds.
Preferably, in the cutting process, the boundary frame of the shrimp larvae at the boundary is assigned to the corresponding sub-image block according to the maximum area principle.
Preferably, after step S23, the method further includes: and visualizing the detection result and the counting result of the shrimp larvae on the image of the shrimp larvae to be detected.
In order to solve the above technical problem, the present invention provides another technical solution as follows: a shrimp larvae counting system based on high resolution and target detection is characterized in that: the system comprises a high-resolution industrial camera, a boundary marking module, a training module, a detection module and a counting module;
the industrial camera is used for acquiring shrimp larvae images used for model training or shrimp larvae images to be detected;
the boundary labeling module is used for carrying out boundary frame labeling on the shrimp larva image to obtain a JSON tag file;
the training module is used for inputting the shrimp larvae images used for model training and the JSON label file into the high-resolution shrimp larvae detection model for training to obtain a trained high-resolution shrimp larvae detection model;
the detection module is used for inputting the shrimp larvae image to be detected into the trained high-resolution shrimp larvae detection model to perform target detection on the shrimp larvae so as to obtain the detection result of the shrimp larvae;
the counting module is used for counting the detection result of the shrimp seeds to obtain the counting result of the shrimp seeds.
Preferably, the shrimp larvae counting system also comprises a water tank for containing the shrimp larvae; the industrial camera is arranged above the water tank in a lifting manner.
(III) advantageous effects
Compared with the prior art, the invention provides a shrimp fry counting method and system based on high resolution and target detection, which have the following beneficial effects: the invention adopts a method based on high resolution and target detection to obtain the accurate positioning and boundary information of the shrimp larvae and further obtain the shrimp larvae counting result, and the industrial camera can acquire the shrimp larvae image with high resolution to be beneficial to the detection of a detection model, so that the method can be suitable for the detection and counting of the shrimp larvae of different varieties including the shrimp larvae of the penaeus monodon and improve the accuracy of the detection and counting; in addition, the shrimp larvae image with high resolution enables a large number of shrimp larvae to be counted at one time so as to cope with the situation that the density of the shrimp larvae is high, and therefore counting efficiency is improved.
Drawings
FIG. 1 is a flow chart of the steps of a model training phase of a shrimp fry counting method based on high resolution and target detection according to the present invention;
FIG. 2 is a flow chart of the steps of the inference use phase of the shrimp fry counting method based on high resolution and target detection according to the invention;
FIG. 3 is a schematic diagram of the high resolution shrimp fry image cutting process of the present invention;
FIG. 4 is a diagram illustrating the maximum area rule for determining the home block of a bounding box according to the present invention;
FIG. 5 is a schematic diagram of detecting a candidate block in substep S222 according to the present invention;
FIG. 6 is a block diagram of a DarkNet41 image block classification network of the present invention;
fig. 7 is a structural view of an industrial camera and a water tank of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a shrimp larvae counting method based on high resolution and target detection, which comprises a model training stage and an inference using stage, wherein the model training stage comprises the following steps of S11-13:
s11: shrimp larvae images for model training were acquired with a high resolution industrial camera.
The industrial camera is preferably on the 4800 million pixel level, and can shoot an ultra-clear image with 8000x6000 resolution; the shrimp larvae image collected by the industrial camera is specifically an RGB image.
S12: and marking a bounding box of the shrimp larva image to obtain a JSON tag file.
For a large number of shrimp larvae images acquired in the model training stage, specifically, an open source labeling tool cvat can be used for labeling the boundary frames of the shrimp larvae images and exporting JSON label files; the bounding box is specifically a bounding rectangle box.
Preferably, after step S12, the method further includes: and cutting the shrimp larva image to obtain a plurality of sub image blocks. For example, the shrimp fry image may be cut into 48 blocks, and the original image may be cut into 48 sub-image blocks with a resolution of 8000x6000 to a resolution of 1000x 1000. It should be understood that the cutting process for the shrimp fry image refers to cropping the picture instead of scaling the compressed picture to reduce the loss of picture information.
In addition, in the cutting process, the boundary frame of the shrimp fry at the boundary can be attributed to the corresponding sub-image block according to the maximum area principle, and the coordinates of the boundary frame (rectangular frame) in the JSON tag file are modified correspondingly. The above maximum area principle is: if the label frame intersects with the boundary of the sub-image block, the attribution of the boundary frame is judged according to the area of each block in the sub-image block after the boundary frame is cut into blocks, and only the boundary frame with the largest area is reserved, for example, as shown in fig. 4.
S13: and inputting the shrimp larva images and the JSON label file into the high-resolution shrimp larva detection model for training to obtain the trained high-resolution shrimp larva detection model.
After the high-resolution shrimp image is cut, the step S13 is specifically: and inputting the plurality of sub image blocks and the JSON label file into a high-resolution shrimp fry detection model for training to obtain a trained high-resolution shrimp fry detection model.
Preferably, the method for obtaining the trained high-resolution shrimp larvae detection model further comprises the following steps: and testing the high-resolution shrimp larvae detection model. Specifically, the sub-image block data and the converted JSON tag file can be converted according to the following steps of 8:2, dividing the ratio into a training set and a test set, wherein the training set is used for training a high-resolution shrimp larva detection model, and the test set is used for testing the model; of course, a data set scale may also be used without undue limitation.
The inference use phase comprises the following steps S21-23:
s21: and acquiring the image of the shrimp larvae to be detected by using an industrial camera.
Preferably, before step S22, the method further includes: the shrimp fry image collected in step S21 is cut as described above to obtain a plurality of subimage blocks.
S22: and inputting the shrimp larvae image into a trained high-resolution shrimp larvae detection model to perform target detection on the shrimp larvae so as to obtain a shrimp larvae detection result.
The step S22 specifically includes the following substeps:
s221: and inputting the plurality of sub-image blocks into the trained high-resolution shrimp larvae detection model.
S222: and classifying the plurality of sub image blocks to extract a plurality of shrimp fry detection candidate image blocks.
S223: and carrying out the target detection of the shrimp larvae on the plurality of shrimp larvae detection candidate image blocks to correspondingly obtain the detection results of the plurality of shrimp larvae.
The High-resolution shrimp larvae detection model is HRSD (High-resolution shrimp detection net), and the HRSD model mainly comprises two sub-networks: darkNet41 and YOLO _ HR. The DarkNet41, i.e. the image block classification network, is responsible for executing the above sub-step S222: for example, 48 sub-images cut out from 4800 million pixels (8000 × 6000 resolution) are classified to extract shrimp fry detection candidate image blocks. For the shrimp larvae image with high resolution, if the shrimp larvae image is directly downsampled and reduced in resolution and then input into a network, tiny shrimp larvae targets disappear, and the detection and counting effects of the shrimp larvae targets are seriously influenced. The invention directly utilizes the high-resolution picture to detect the shrimp larvae, however, after the high-resolution picture is divided into blocks, the shrimp larvae are possibly distributed unevenly during the collection, and the shrimp larvae targets are not all in all the image blocks, which is particularly obvious in the high-resolution picture. Therefore, whether a target exists is judged by classifying the sub image blocks through the DarkNet41, as shown in FIG. 5, if the target exists, the sub image block is a shrimp fry detection candidate image block, and if the target does not exist, the next detection is not performed, so that only the sub image blocks with the targets possibly existing are detected subsequently, and the detection efficiency can be improved.
The DarkNet41 is modified based on DarkNet53, and the DarkNet41 is constructed for classifying the detection candidate blocks. The DarkNet41 reduces the number of convolution layers, respectively reduces the number of 5 stages of Residual Block from [1,2,8, 4] to [1,2,6, 4], the structure of the DarkNet41 is shown in FIG. 6, which comprises a total of 41 convolution layers, and sub-image blocks with 1000 × 1000 resolution are downsampled to 256 × 256 resolution and input into the DarkNet41 for classification.
Also, YOLO _ HR, i.e., the shrimp larvae detection network, is responsible for performing the above-described substep S223. According to the YOLO _ HR disclosed by the invention, on the basis of yolov4, the original DarkNet53 backbone network is replaced by the efficient RepVGGA1 backbone network to construct the shrimp larvae detection network, so that the shrimp larvae detection performance is improved. The speed of the RepVGG backbone network is obviously improved after hardware optimization only by the combination of 3 multiplied by 3 convolution and ReLU; the RepVGGA1 is the A1 version of the RepVGG, the RepVGG and the DarkNet53 have 5 stages, the number of CR modules in each stage is {1,2,4,14,1}, each CR module uses a CBR module to reduce the resolution of a feature map before, the feature map output in the 5 th stage is enhanced by an SPP module, then is fused with the feature maps output in the third and fourth stages by a PANET structure, and then is analyzed by a head structure which is the same as the YOLOv3, so that a final detection result is obtained.
S23: and counting the detection result of the shrimp seeds to obtain the counting result of the shrimp seeds.
On the basis of the sub-step S223, the step S23 specifically includes: and counting and combining the detection results of the plurality of young shrimps to obtain the counting result of the young shrimps. It should be understood that the detection result of the shrimp larvae detects all the shrimp larvae on the shrimp larvae detection candidate image block. For example, for 48 sub image blocks, 38 of the image blocks are shrimp larvae detection candidate image blocks, further, 2 shrimp larvae targets are detected and counted in the first shrimp larvae detection candidate image block, 4 shrimp larvae targets are detected and counted in the second shrimp larvae detection candidate image block, and finally, the shrimp larvae counts of the 38 shrimp larvae detection candidate image blocks are combined to obtain the counting result of the shrimp larvae image to be detected.
Further, it is preferable that, after step S23, the method further includes: the detection result and the counting result of the shrimp larvae are visualized on the shrimp larvae image to be detected, namely, the detected targets of the shrimp larvae and the counting result of the shrimp larvae are displayed on the shrimp larvae image, so that the detection counting result can be conveniently and clearly understood.
The invention also provides a shrimp larvae counting system based on high resolution and target detection, which comprises a high resolution industrial camera, a boundary marking module, a training module, a detection module and a counting module.
The industrial camera is used for collecting shrimp larvae images used for model training or shrimp larvae images to be detected.
And the boundary labeling module is used for carrying out boundary frame labeling on the shrimp larva image to obtain a JSON tag file.
And the training module is used for inputting the shrimp larvae images used for model training and the JSON label file into the high-resolution shrimp larvae detection model for training so as to obtain the trained high-resolution shrimp larvae detection model.
The detection module is used for inputting the shrimp larvae images to be detected into the trained high-resolution shrimp larvae detection model to perform target detection on the shrimp larvae so as to obtain the detection result of the shrimp larvae.
The counting module is used for counting the detection result of the shrimp seeds to obtain the counting result of the shrimp seeds.
For the specific functions of the shrimp larvae counting system of the present invention, reference may be made to the above-mentioned descriptions of the shrimp larvae counting method, and no redundant description is provided herein.
In addition, referring to fig. 7, the shrimp larvae counting system may further include a water tank 20 for containing the shrimp larvae; the water tank 20 is preferably a white plastic water tank, and the water tank can be 1300x900x150mm in size.
Preferably, the industrial camera 10 is arranged above the water tank 20 in a liftable manner, the industrial camera 10 is specifically lifted through the telescopic adjusting device, and the industrial camera 10 has the characteristics of high image stability, high transmission capability, high anti-interference capability and the like, and is far superior to a common digital camera in the aspects of shooting speed, precision, repeatability and the like.
When acquiring the shrimp larvae image, firstly, pouring the shrimp larvae to be counted (20000 to 50000 tails are good) into a white plastic water tank, and shaking the shrimp larvae uniformly, wherein the water level is about 1 cm. Adjusting the height of the industrial camera to enable the water surface to occupy more than 95% of the visual field of the picture, and then shooting the shrimp fry picture; in the early stage, a large number of pictures need to be collected in order to train a high-resolution shrimp larva detection model; in the later period, when the system is used in reasoning, each water tank only needs to shoot 5-10 pieces of water.
Compared with the prior art, the invention provides a shrimp fry counting method and system based on high resolution and target detection, which have the following beneficial effects: (1) The invention adopts a method based on high resolution and target detection to obtain the accurate positioning and boundary information of the shrimp larvae and further obtain the shrimp larvae counting result, and the industrial camera can acquire the shrimp larvae image with high resolution to be beneficial to the detection of a detection model, so that the method can be suitable for the detection and counting of the shrimp larvae of different varieties including the shrimp larvae of the penaeus monodon and improve the accuracy of the detection and counting; (2) In addition, the high-resolution shrimp larvae images enable a large number of shrimp larvae to be counted at one time so as to cope with the situation that the density of the shrimp larvae is high, and therefore counting efficiency is improved; (3) The high-resolution parallel network HRSD suitable for shrimp fry counting is built based on yolov4, a plurality of sub-image blocks (such as 48 images) are used in parallel in training and reasoning, the counting speed is increased, and the counting accuracy is improved; (4) The water tank can accommodate a large number of shrimp seeds at one time, and the counting efficiency is improved; the requirements of shrimp larvae on the background and light are not high, and extra equipment is not needed for light supplement when shrimp larvae images are shot, or containers are poured backwards for many times, so that the method is convenient and rapid, and the influence on the shrimp larvae is extremely small.
It should be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of additional like elements in a process, method, article, or apparatus that comprises the element.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (10)
1. The shrimp fry counting method based on high resolution and target detection is characterized by comprising a model training stage and an inference use stage, wherein the model training stage comprises the following steps:
s11: acquiring shrimp larva images for model training by using a high-resolution industrial camera;
s12: carrying out boundary frame marking on the shrimp larva image to obtain a JSON tag file;
s13: inputting the shrimp fry image and the JSON label file into a high-resolution shrimp fry detection model for training to obtain a trained high-resolution shrimp fry detection model;
the inference use phase comprises the following steps:
s21: acquiring a shrimp larva image to be detected by using the industrial camera;
s22: inputting the shrimp larvae image into a trained high-resolution shrimp larvae detection model to perform target detection on the shrimp larvae so as to obtain a shrimp larvae detection result;
s23: and counting the detection result of the shrimp larvae to obtain the counting result of the shrimp larvae.
2. The shrimp larvae counting method as claimed in claim 1, further comprising, after step S12: and cutting the shrimp larva image to obtain a plurality of sub image blocks.
3. The shrimp larvae counting method as claimed in claim 2, wherein the step S13 is specifically: and inputting the plurality of sub-image blocks and the JSON tag file into a high-resolution shrimp larvae detection model for training to obtain a trained high-resolution shrimp larvae detection model.
4. The shrimp larvae counting method as claimed in claim 2, wherein: before step S22, the method further includes: and performing the cutting processing on the shrimp fry image acquired in the step S21 to obtain the plurality of sub image blocks.
5. The shrimp larvae counting method as claimed in claim 4, wherein the step S22 comprises the following substeps:
s221: inputting the plurality of sub-image blocks into a trained high-resolution shrimp larvae detection model;
s222: classifying the plurality of sub image blocks to extract a plurality of shrimp fry detection candidate image blocks;
s223: and carrying out the target detection of the shrimp larvae on the plurality of shrimp larvae detection candidate image blocks to correspondingly obtain the detection results of the plurality of shrimp larvae.
6. The shrimp larvae counting method according to claim 5, wherein the step S23 comprises: and counting and combining the detection results of the plurality of shrimp seeds to obtain the counting result of the shrimp seeds.
7. The shrimp fry counting method as claimed in claim 2, wherein in the cutting process, the shrimp fry bounding boxes at the boundary are classified into the corresponding subimage blocks according to a maximum area principle.
8. The shrimp larvae counting method as claimed in claim 1, further comprising, after step S23: and visualizing the detection result and the counting result of the shrimp larvae on the image of the shrimp larvae to be detected.
9. A shrimp larvae counting system based on high resolution and target detection is characterized in that: the system comprises a high-resolution industrial camera, a boundary marking module, a training module, a detection module and a counting module;
the industrial camera is used for acquiring shrimp larvae images used for model training or shrimp larvae images to be detected;
the boundary labeling module is used for carrying out boundary frame labeling on the shrimp fry image to obtain a JSON tag file;
the training module is used for inputting the shrimp larvae images used for model training and the JSON label file into the high-resolution shrimp larvae detection model for training to obtain a trained high-resolution shrimp larvae detection model;
the detection module is used for inputting the shrimp larvae images to be detected into a trained high-resolution shrimp larvae detection model to perform target detection on the shrimp larvae so as to obtain the detection result of the shrimp larvae;
the counting module is used for counting the detection result of the shrimp seeds to obtain the counting result of the shrimp seeds.
10. The shrimp fry counting system of claim 9, wherein: the water tank is used for containing the shrimp larvae; the industrial camera is arranged above the water tank in a liftable manner.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211666456.2A CN115937169A (en) | 2022-12-23 | 2022-12-23 | Shrimp fry counting method and system based on high resolution and target detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211666456.2A CN115937169A (en) | 2022-12-23 | 2022-12-23 | Shrimp fry counting method and system based on high resolution and target detection |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115937169A true CN115937169A (en) | 2023-04-07 |
Family
ID=86650838
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211666456.2A Pending CN115937169A (en) | 2022-12-23 | 2022-12-23 | Shrimp fry counting method and system based on high resolution and target detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115937169A (en) |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875825A (en) * | 2018-06-12 | 2018-11-23 | 河南工业大学 | A kind of Grainhouse injurious insect detection method based on image block |
| CN111161246A (en) * | 2019-12-30 | 2020-05-15 | 歌尔股份有限公司 | Product defect detection method, device and system |
| CN111814673A (en) * | 2020-07-08 | 2020-10-23 | 重庆农村商业银行股份有限公司 | Method, device and equipment for correcting text detection bounding box and storage medium |
| CN112861767A (en) * | 2021-02-26 | 2021-05-28 | 北京农业信息技术研究中心 | Small-volume pest detection method and system on pest sticking plate image |
| WO2021179491A1 (en) * | 2020-03-13 | 2021-09-16 | 平安科技（深圳）有限公司 | Image processing method and apparatus, computer device and storage medium |
| CN113762220A (en) * | 2021-11-03 | 2021-12-07 | 通号通信信息集团有限公司 | Object recognition method, electronic device, and computer-readable storage medium |
| CN114066842A (en) * | 2021-11-12 | 2022-02-18 | 浙江托普云农科技股份有限公司 | Method, system and device for counting number of ears and storage medium |
| CN114782368A (en) * | 2022-04-25 | 2022-07-22 | 华南农业大学 | Shrimp larva automatic counting method and system based on visual detection and fusion of multiple dynamic thresholds |
| CN115147347A (en) * | 2022-04-22 | 2022-10-04 | 宁波大学 | Method for detecting surface defects of malleable cast iron pipe fitting facing edge calculation |
| CN115170571A (en) * | 2022-09-07 | 2022-10-11 | 赛维森(广州)医疗科技服务有限公司 | Method and device for identifying pathological images of hydrothorax and ascites cells and medium |
| CN115222666A (en) * | 2022-06-14 | 2022-10-21 | 华南农业大学 | Shrimp larvae counting method, system, equipment and storage medium based on key point detection |
-
2022
- 2022-12-23 CN CN202211666456.2A patent/CN115937169A/en active Pending
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108875825A (en) * | 2018-06-12 | 2018-11-23 | 河南工业大学 | A kind of Grainhouse injurious insect detection method based on image block |
| CN111161246A (en) * | 2019-12-30 | 2020-05-15 | 歌尔股份有限公司 | Product defect detection method, device and system |
| WO2021179491A1 (en) * | 2020-03-13 | 2021-09-16 | 平安科技（深圳）有限公司 | Image processing method and apparatus, computer device and storage medium |
| CN111814673A (en) * | 2020-07-08 | 2020-10-23 | 重庆农村商业银行股份有限公司 | Method, device and equipment for correcting text detection bounding box and storage medium |
| CN112861767A (en) * | 2021-02-26 | 2021-05-28 | 北京农业信息技术研究中心 | Small-volume pest detection method and system on pest sticking plate image |
| CN113762220A (en) * | 2021-11-03 | 2021-12-07 | 通号通信信息集团有限公司 | Object recognition method, electronic device, and computer-readable storage medium |
| CN114066842A (en) * | 2021-11-12 | 2022-02-18 | 浙江托普云农科技股份有限公司 | Method, system and device for counting number of ears and storage medium |
| CN115147347A (en) * | 2022-04-22 | 2022-10-04 | 宁波大学 | Method for detecting surface defects of malleable cast iron pipe fitting facing edge calculation |
| CN114782368A (en) * | 2022-04-25 | 2022-07-22 | 华南农业大学 | Shrimp larva automatic counting method and system based on visual detection and fusion of multiple dynamic thresholds |
| CN115222666A (en) * | 2022-06-14 | 2022-10-21 | 华南农业大学 | Shrimp larvae counting method, system, equipment and storage medium based on key point detection |
| CN115170571A (en) * | 2022-09-07 | 2022-10-11 | 赛维森(广州)医疗科技服务有限公司 | Method and device for identifying pathological images of hydrothorax and ascites cells and medium |
Non-Patent Citations (1)
| Title |
|---|
| 张丹丹: "《基于深度神经网络技术的高分遥感图像处理及应用》", 31 August 2020, 中国宇航出版社 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109145830B (en) | Intelligent water gauge identification method | |
| CN108686978B (en) | ARM-based fruit category and color sorting method and system | |
| CN110059558A (en) | A kind of orchard barrier real-time detection method based on improvement SSD network | |
| CN108961229A (en) | Cardiovascular OCT image based on deep learning easily loses plaque detection method and system | |
| CN108334847A (en) | A kind of face identification method based on deep learning under real scene | |
| CN111401353B (en) | Method, device and equipment for identifying mathematical formula | |
| CN110276386A (en) | A kind of apple grading method and system based on machine vision | |
| CN112801230B (en) | Intelligent acceptance method for unmanned aerial vehicle of power distribution line | |
| CN113392775A (en) | Sugarcane seedling automatic identification and counting method based on deep neural network | |
| CN111626993A (en) | Image automatic detection counting method and system based on embedded FEFnet network | |
| CN109886978B (en) | End-to-end alarm information identification method based on deep learning | |
| CN112069985B (en) | High-resolution field image rice spike detection and counting method based on deep learning | |
| CN113160062B (en) | Infrared image target detection method, device, equipment and storage medium | |
| CN112967255A (en) | Shield segment defect type identification and positioning system and method based on deep learning | |
| CN110570427A (en) | Remote sensing image semantic segmentation method and device fusing edge detection | |
| CN110287806A (en) | A kind of traffic sign recognition method based on improvement SSD network | |
| CN114140665A (en) | Dense small target detection method based on improved YOLOv5 | |
| CN111339902A (en) | Liquid crystal display number identification method and device of digital display instrument | |
| CN111507353B (en) | Chinese field detection method and system based on character recognition | |
| CN115861409B (en) | Soybean leaf area measuring and calculating method, system, computer equipment and storage medium | |
| CN112163508A (en) | Character recognition method and system based on real scene and OCR terminal | |
| CN114359235A (en) | Wood surface defect detection method based on improved YOLOv5l network | |
| CN114937266A (en) | Hard shell clam biological sign identification method based on YOLOX-S | |
| CN116052105A (en) | Pavement crack identification classification and area calculation method, system, equipment and terminal | |
| CN113763196A (en) | Orchard yield measuring system based on improved YOLOv3 |
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 |