CN116824369A - Litchi diseases real-time detection system and method based on edge calculation and yolov7-tiny - Google Patents
Litchi diseases real-time detection system and method based on edge calculation and yolov7-tiny Download PDFInfo
- Publication number
- CN116824369A CN116824369A CN202310721957.4A CN202310721957A CN116824369A CN 116824369 A CN116824369 A CN 116824369A CN 202310721957 A CN202310721957 A CN 202310721957A CN 116824369 A CN116824369 A CN 116824369A
- Authority
- CN
- China
- Prior art keywords
- module
- map
- tiny
- average
- yolov7
- 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
- 238000004364 calculation method Methods 0.000 title claims abstract description 30
- 201000010099 disease Diseases 0.000 title claims abstract description 29
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 title claims abstract description 29
- 241001629511 Litchi Species 0.000 title claims abstract description 18
- 238000000034 method Methods 0.000 title claims abstract description 16
- 238000011897 real-time detection Methods 0.000 title claims abstract description 14
- 238000001514 detection method Methods 0.000 claims abstract description 37
- 238000004891 communication Methods 0.000 claims abstract description 5
- 238000013528 artificial neural network Methods 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 9
- 238000013527 convolutional neural network Methods 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 4
- 230000003044 adaptive effect Effects 0.000 claims description 3
- 230000006872 improvement Effects 0.000 claims description 3
- 230000003993 interaction Effects 0.000 claims description 3
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 244000183278 Nephelium litchi Species 0.000 claims 4
- 235000015742 Nephelium litchi Nutrition 0.000 claims 4
- 238000005516 engineering process Methods 0.000 abstract description 6
- 230000005540 biological transmission Effects 0.000 abstract description 2
- 238000003973 irrigation Methods 0.000 description 3
- 230000002262 irrigation Effects 0.000 description 3
- 241000196324 Embryophyta Species 0.000 description 2
- 241000607479 Yersinia pestis Species 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000004720 fertilization Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 1
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
- G06V20/188—Vegetation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Multimedia (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Mathematical Optimization (AREA)
- Pure & Applied Mathematics (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Computational Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Operations Research (AREA)
- Probability & Statistics with Applications (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Algebra (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a litchi disease real-time detection system and a method based on edge calculation and yolov7-tiny, wherein the litchi disease real-time detection system comprises an image recognition module, a yolov7-tiny module, an attention CBAM module, a mAP module and an edge deployment module, wherein the image recognition module is used for receiving influence data transmitted by a shooting device and forwarding the data, the mAP module is used for comprehensively measuring a detection result, the calculation mode is that mAP=average precision summation of all categories is divided by all categories, the attention CBAM module is used for the yolov7-tiny module, a model is made to grasp key points in a neural network, and the feature scaling module is used for scaling image features; the litchi disease detection method is a set of high-technology litchi disease condition collecting and detecting information transmission device integrating computer remote communication, computer hardware technology and multimedia technology, can judge the litchi disease condition in real time, simplifies detection steps and greatly improves detection efficiency.
Description
Technical Field
The invention relates to the field of pest and disease detection, in particular to a litchi disease real-time detection system and method based on edge calculation and yolov 7-tiny.
Background
At present, most litchi gardens in China still adopt a traditional manual mode for management, and irrigation of fruit trees, fertilization of fruit trees, disease and pest prediction, prevention and treatment and the like are all judged and decided by human experience. The manual management mode is rough and has low purposefulness, and particularly, the traditional irrigation modes such as flood irrigation and the like are adopted, so that serious water resource waste is caused.
The existing disease detection is usually carried out in a manual detection mode, the detection efficiency is low, the damage state of plants can be judged only through the appearance characteristics of the diseases in the traditional detection, and the detection result precision is poor.
Disclosure of Invention
The technical problems to be solved by the invention are as follows: the existing disease detection is usually carried out in a manual detection mode, the detection efficiency is low, the damage state of plants can be judged only through the appearance characteristics of the diseases in the traditional detection, and the detection result precision is poor.
The invention solves the technical problems through the following technical scheme, and the litchi disease real-time detection system based on edge calculation and yolov7-tiny comprises an image recognition module, a yolov7-tiny module, an attention CBAM module, a mAP module and an edge deployment module;
the image recognition module is used for receiving the influence data transmitted by the shooting device and forwarding the data;
the mAP module is used for comprehensively measuring the detection result, and the calculation mode is mAP=average precision summation of all categories divided by all categories;
the attention CBAM module is used for a yolov7-tiny module and enables a model to grasp important points in a neural network;
the feature scaling module is used for scaling the image features;
the improved Y0L0v7-tiny module is used for learning data which is not successfully detected, and the learning method is based on network deep learning of Y0L0v 7;
the edge deployment module is used for deploying storage weight parameters to judge new disease image data;
the image recognition module, the mAP module, the improved Y0L0v7-tiny module, the edge deployment module and the interaction module are sequentially in communication connection.
Preferably, the specific processing steps of the mAP module are as follows:
and (S1) S: the mAP module receives the database and compares the database with data stored in the model database, and the comparison results comprise four types, namely TP, FP, FN, TN, wherein TP is judged to be positive in the positive type, FP is judged to be negative in the negative type, FN is judged to be negative in the positive type, and TN is judged to be negative in the negative type;
s2: calculating accuracy and recall rate;
the calculation formula of the accuracy is as follows: precision = TP/(tp+fp)
The calculation formula of the recall rate is as follows: recall=tp/(tp+fn);
s3: calculating average accuracy of each category: average Precision per class = (Precision 1+precision2+, +precision N)/N
Calculating average recall rates of all the categories: average Recall per class = (recal1+recal2+ & gt+recalln)/N
Where N represents the total number of categories, precision1, precision2, precision N represents the Precision of each category, recall1, recall2, recall N represents the Recall of each category, respectively;
s4: calculate all class average accuracy Mean Average Precision = (Average Precision1+ Average Precision2 +2..+ Average PrecisionN)/N
Calculate Average Recall for all categories Mean Average Recall = (Average recal1+average recal2+, +average RecallN)/N
Where N represents the total number of categories, average Precision1, average Precision2,.. Average PrecisionN represent the Average accuracy of each category, respectively, average Recall1, average Recall2,..and Average Recall N represent the Average Recall of each category;
s5: the average value mAP of each category is obtained after the calculation, and the closer the mAP is to 1, the higher the detection accuracy is, and the lower the detection accuracy is otherwise.
Preferably, the scaling method of the feature scaling module is mean normalization, and the feature scaling formula is as follows:
a＝(x-μ)/σ
where a is the value normalized by the average, x represents the original value, μ represents the average value, and σ represents the standard deviation.
Preferably, the attention module specifically works as follows:
given an intermediate feature map, the CBAM module would infer the attention map sequentially along two independent dimensions (channels and spaces) and then multiply the attention map with the input feature map for adaptive feature optimization, since CBAM is a lightweight generic module, negligible overhead of this module would seamlessly integrate it into any CNN architecture and could perform end-to-end training with the underlying CNN.
Preferably, the edge deployment module adopts Jetson nano equipment, and Jetson nano has a camera interface, and the Jetson nano is connected with the nano camera through the camera interface to acquire images.
Preferably, the edge deployment module embeds an identification program, and the program includes a trained yolov7tiny model.
The litchi diseases real-time detection method based on edge calculation and yolov7-tiny specifically comprises the following steps:
step one: constructing an improved yolov7-tiny model, wherein the model improvement module comprises a convolution attention mechanism module (CBAM) and a feature scaling module, and training the yolov7-tiny model by using a training set to obtain a final trained model;
step two: deploying the trained yolov7-tiny model into an edge deployment module equipment end Jetson nano;
step three: the image recognition module, namely a nano camera, receives a real-time disease video stream, wherein the video stream comprises a plurality of continuous images;
step four: and inputting the video stream into external equipment to a final target detection model deployed on an Injeida nano development board for target detection, and outputting a target detection result.
Compared with the prior art, the invention has the following advantages:
the deep learning in the disease detection process is performed by using a deep network learning architecture of Y0L0v7-tiny in the prior art, so that the learning effect is stronger, the confidence is higher, and the mAP calculation can be more accurate;
by setting the feature scaling module, the features can be scaled and normalized, so that the features can be evenly amplified or reduced, further, the feature comparison result is more accurate in the mAP calculation process, meanwhile, the feature scaling module can be learned and stored in a model library by Y0L0v7tiny after scaling the features, thus, images with the same features are encountered later, the calculation result is faster,
the litchi disease detection method is a set of high-technology litchi disease condition collecting and detecting information transmission device integrating computer remote communication, computer hardware technology and multimedia technology, can judge the litchi disease condition in real time, simplifies detection steps and greatly improves detection efficiency.
Drawings
Fig. 1 is a system block diagram of the present invention.
Detailed Description
The following describes in detail the examples of the present invention, which are implemented on the premise of the technical solution of the present invention, and detailed embodiments and specific operation procedures are given, but the scope of protection of the present invention is not limited to the following examples.
As shown in fig. 1, this embodiment provides a technical solution: the litchi damage real-time detection system based on edge calculation and yolov7-tiny comprises an image recognition module, a yolov7-tiny module, a attention CBAM module, a mAP module and an edge deployment module;
the image recognition module is used for receiving the influence data transmitted by the shooting device and forwarding the data;
the mAP module is used for comprehensively measuring the detection result, and the calculation mode is mAP=average precision summation of all the categories divided by all the categories;
the attention CBAM module is used for a yolov7-tiny module, and the model is made to grasp the key points in the neural network;
the feature scaling module is used for scaling the image features;
the improved Y0L0v7-tiny module is used for learning data which is not successfully detected, and the learning method is based on network deep learning of Y0L0v 7;
the edge deployment module is used for deploying and storing weight parameters to judge new disease image data;
the image recognition module, the mAP module, the improved Y0L0v7-tiny module, the edge deployment module and the interaction module are sequentially in communication connection.
The mAP module comprises the following specific processing steps:
and (S1) S: the mAP module receives the database and compares the database with data stored in the model database, and the comparison results comprise four types, namely TP, FP, FN, TN, wherein TP is judged to be positive in the positive type, FP is judged to be negative in the negative type, FN is judged to be negative in the positive type, and TN is judged to be negative in the negative type;
s2: calculating accuracy and recall rate;
the calculation formula of the accuracy is as follows: precision = TP/(tp+fp)
The calculation formula of the recall rate is as follows: recall=tp/(tp+fn);
s3: calculating average accuracy of each category: average Precision per class = (Precision 1+precision2+, +precision N)/N
Calculating average recall rates of all the categories: average Recall per class = (recal1+recal2+ & gt+recalln)/N
Where N represents the total number of categories, precision1, precision2, precision N represents the Precision of each category, recall1, recall2, recall N represents the Recall of each category, respectively;
s4: calculate all class average accuracy Mean Average Precision = (Average Precision1+ Average Precision2 +2..+ Average PrecisionN)/N
Calculate Average Recall for all categories Mean Average Recall = (Average recal1+average recal2+, +average RecallN)/N
Where N represents the total number of categories, average Precision1, average Precision2,.. Average PrecisionN represent the Average accuracy of each category, respectively, average Recall1, average Recall2,..and Average Recall N represent the Average Recall of each category;
s5: the average value mAP of each category is obtained after the calculation, and the closer the mAP is to 1, the higher the detection accuracy is, and the lower the detection accuracy is otherwise.
The scaling method of the feature scaling module is mean normalization, and the feature scaling formula is as follows:
a＝(x-μ)/σ
where a is the value normalized by the average, x represents the original value, μ represents the average value, and σ represents the standard deviation.
The attention module works specifically as follows:
given an intermediate feature map, the CBAM module would infer the attention map sequentially along two independent dimensions (channels and spaces) and then multiply the attention map with the input feature map for adaptive feature optimization, since CBAM is a lightweight generic module, negligible overhead of this module would seamlessly integrate it into any CNN architecture and could perform end-to-end training with the underlying CNN.
The edge deployment module adopts Jetson nano equipment, and Jetson nano has a camera interface, and a nano camera is connected through the camera interface to acquire images.
The edge deployment module embeds an identification program that includes a trained yolov7tiny model.
The litchi diseases real-time detection method based on edge calculation and yolov7-tiny specifically comprises the following steps:
step one: constructing an improved yolov7-tiny model, wherein the model improvement module comprises a convolution attention mechanism module (CBAM) and a feature scaling module, and training the yolov7-tiny model by using a training set to obtain a final trained model;
step two: deploying the trained yolov7-tiny model into an edge deployment module equipment end Jetson nano;
step three: the image recognition module, namely a nano camera, receives a real-time disease video stream, wherein the video stream comprises a plurality of continuous images;
step four: inputting the video stream into external equipment to a final target detection model deployed on an Injeida nano development board for target detection, and outputting a target detection result.
While embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the invention, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the invention.
Claims (6)
1. The litchi disease real-time detection system based on edge calculation and yolov7-tiny is characterized by comprising an image recognition module, a yolov7-tiny module, a attention CBAM module, a mAP module and an edge deployment module;
the image recognition module is used for receiving the influence data transmitted by the shooting device and forwarding the data;
the mAP module is used for comprehensively measuring the detection result, and the calculation mode is mAP=average precision summation of all categories divided by all categories;
the attention CBAM module is used for a yolov7-tiny module and enables a model to grasp important points in a neural network;
the feature scaling module is used for scaling the image features;
the improved Y0L0v7-tiny module is used for learning data which is not successfully detected, and the learning method is based on network deep learning of Y0L0v 7;
the edge deployment module is used for deploying storage weight parameters to judge new disease image data;
the image recognition module, the mAP module, the improved Y0L0v7-tiny module, the edge deployment module and the interaction module are sequentially in communication connection.
2. The litchi chinensis real-time detection system based on edge calculation and yolov 7-tini according to claim 1, wherein the system is characterized in that: the mAP module comprises the following specific processing steps:
and (S1) S: the mAP module receives the database and compares the database with data stored in the model database, and the comparison results comprise four types, namely TP, FP, FN, TN, wherein TP is judged to be positive in the positive type, FP is judged to be negative in the negative type, FN is judged to be negative in the positive type, and TN is judged to be negative in the negative type;
s2: calculating accuracy and recall rate;
the calculation formula of the accuracy is as follows: precision = TP/(tp+fp)
The calculation formula of the recall rate is as follows: recall=tp/(tp+fn);
s3: calculating average accuracy of each category: averageprecision inclusion= (Precision 1+precision2+ & gt+precision N)/N
Calculating average recall rates of all the categories: averagerecallperclass= (recal1+recal2+ & gt RecallN)/N
Where N represents the total number of categories, precision1, precision2, precision N represents the Precision of each category, recall1, recall2, recall N represents the Recall of each category, respectively;
s4: calculate all class average accuracy meaneaageprecision= (Average Precision1+averageprecision2+, +averageprecision N)/N
The average recall ratio meaneagerecall= (averagerecall1+averagerecall2+.+ AverageRecallN)/N for all categories was calculated
Where N represents the total number of categories, averagePrecision1, averagePrecision2,..averageprecision N represents the average precision of each category, averageRecall1, averageRecall2,..averagerecall N represents the average recall of each category, respectively.
S5: the average value mAP of each category is obtained after the calculation, and the closer the mAP is to 1, the higher the detection accuracy is, and the lower the detection accuracy is otherwise.
3. The litchi chinensis real-time detection system based on edge calculation and yolov 7-tini according to claim 1, wherein the system is characterized in that: the scaling method of the feature scaling module is mean normalization, and the feature scaling formula is as follows:
a＝(x-μ)/σ
where a is the value normalized by the average, x represents the original value, μ represents the average value, and σ represents the standard deviation.
4. The litchi chinensis real-time detection system based on edge calculation and yolov 7-tini according to claim 1, wherein the system is characterized in that: the attention module works specifically as follows:
given an intermediate feature map, the CBAM module would infer the attention map sequentially along two independent dimensions (channels and spaces) and then multiply the attention map with the input feature map for adaptive feature optimization, since CBAM is a lightweight generic module, negligible overhead of this module would seamlessly integrate it into any CNN architecture and could perform end-to-end training with the underlying CNN.
5. The litchi chinensis real-time detection system based on edge calculation and yolov 7-tini according to claim 1, wherein the system is characterized in that: the edge deployment module is embedded with an identification program, and the program comprises a trained yolov7tiny model.
6. The real-time litchi disease detection method based on edge calculation and yolov7-tiny is characterized by comprising the following steps of: the method specifically comprises the following steps:
step one: constructing an improved yolov7-tiny model, wherein the model improvement module comprises a convolution attention mechanism module (CBAM) and a feature scaling module, and training the yolov7-tiny model by using a training set to obtain a final trained model;
step two: deploying the trained yolov7-tiny model into an edge deployment module equipment end Jetson nano;
step three: the image recognition module, namely a nano camera, receives a real-time disease video stream, wherein the video stream comprises a plurality of continuous images;
step four: and inputting the video stream into external equipment for target detection, and outputting a target detection result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310721957.4A CN116824369A (en) | 2023-06-19 | 2023-06-19 | Litchi diseases real-time detection system and method based on edge calculation and yolov7-tiny |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310721957.4A CN116824369A (en) | 2023-06-19 | 2023-06-19 | Litchi diseases real-time detection system and method based on edge calculation and yolov7-tiny |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116824369A true CN116824369A (en) | 2023-09-29 |
Family
ID=88123493
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310721957.4A Pending CN116824369A (en) | 2023-06-19 | 2023-06-19 | Litchi diseases real-time detection system and method based on edge calculation and yolov7-tiny |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116824369A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117036361A (en) * | 2023-10-10 | 2023-11-10 | 云南大学 | Power grid transmission line smoke detection method, system, electronic equipment and medium |
-
2023
- 2023-06-19 CN CN202310721957.4A patent/CN116824369A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117036361A (en) * | 2023-10-10 | 2023-11-10 | 云南大学 | Power grid transmission line smoke detection method, system, electronic equipment and medium |
| CN117036361B (en) * | 2023-10-10 | 2024-02-20 | 云南大学 | Power grid transmission line smoke detection method, system, electronic equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111611878A (en) | Method for crowd counting and future people flow prediction based on video image | |
| CN106709462A (en) | Indoor positioning method and device | |
| CN112560745B (en) | Method for discriminating personnel on electric power operation site and related device | |
| CN111209832B (en) | Auxiliary obstacle avoidance training method, equipment and medium for substation inspection robot | |
| CN116824369A (en) | Litchi diseases real-time detection system and method based on edge calculation and yolov7-tiny | |
| WO2022048582A1 (en) | Method and device for optical flow information prediction, electronic device, and storage medium | |
| CN110633643A (en) | Abnormal behavior detection method and system for smart community | |
| CN111161315A (en) | Multi-target tracking method and system based on graph neural network | |
| CN113903081A (en) | Visual identification artificial intelligence alarm method and device for images of hydraulic power plant | |
| CN106846378A (en) | Across video camera object matching and tracking that a kind of combination topology of spacetime is estimated | |
| CN112001347A (en) | Motion recognition method based on human skeleton shape and detection target | |
| CN110519582A (en) | A kind of crusing robot data collection system and collecting method | |
| CN108764456A (en) | Airborne target identification model construction platform, airborne target recognition methods and equipment | |
| CN112349057A (en) | Deep learning-based indoor smoke and fire detection method | |
| CN109389156A (en) | A kind of training method, device and the image position method of framing model | |
| CN108537825A (en) | A kind of method for tracking target based on transfer learning Recurrent networks | |
| CN116824626A (en) | Artificial intelligent identification method for abnormal state of animal | |
| CN112766305B (en) | Visual SLAM closed loop detection method based on end-to-end measurement network | |
| CN114170686A (en) | Elbow bending behavior detection method based on human body key points | |
| CN110841143B (en) | Method and system for predicting state of infusion pipeline | |
| CN102497495B (en) | Target association method for multi-camera monitoring system | |
| CN114627496A (en) | Robust pedestrian re-identification method based on depolarization batch normalization of Gaussian process | |
| CN113344968A (en) | Orchard fruit identification and yield statistical system and method | |
| CN113516232A (en) | Training method of neural network model based on self-attention mechanism | |
| CN112257568B (en) | Intelligent real-time supervision and error correction system and method for individual soldier queue actions |
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 |