CN113177486A - Dragonfly order insect identification method based on regional suggestion network - Google Patents
Dragonfly order insect identification method based on regional suggestion network Download PDFInfo
- Publication number
- CN113177486A CN113177486A CN202110480792.7A CN202110480792A CN113177486A CN 113177486 A CN113177486 A CN 113177486A CN 202110480792 A CN202110480792 A CN 202110480792A CN 113177486 A CN113177486 A CN 113177486A
- Authority
- CN
- China
- Prior art keywords
- network
- network model
- dragonfly
- training
- data set
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a dragonfly order insect identification method based on a regional suggestion network, which comprises the following steps: s1, cleaning and sorting images of dragonfly order insects to obtain a data set of the dragonfly order insects; s2, enhancing the data set of the dragonfly order insect image to obtain an enhanced data set; s3, dividing the enhanced data set to obtain a training set, a verification set and a test set of the dragonfly order insect images; s4, constructing a deep convolutional network model based on the regional suggestion network; s5, training a deep convolution network model by using the training set and the verification set to obtain a trained network model; and S6, inputting the test set into the trained network model, and outputting to obtain a classification result of the test set. The method can enable the identification processing to be simple and rapid, save a large amount of labor cost, and solve the problem of difficult identification caused by complex background of the dragonfly-order insect pictures in natural environment.
Description
Technical Field
The invention relates to the field of identification, in particular to a dragonfly order insect identification method based on a regional suggestion network.
Background
The existing dragonfly order automatic identification algorithm takes manually designed characteristics as a classification basis, and a traditional identification mode is used for constructing an identification frame, so that only a plurality of sample pictures of dragonflies can be identified, the identification rate is low, and the identification capability of the dragonflies with complex backgrounds in natural environment is not provided;
existing automatic insect identification algorithms are often implemented in two steps. In the first step, detection is carried out. The detection algorithm is firstly realized for the target to be identified, and the process depends on a large amount of manual marking information, so that the time and the labor are consumed, and the early cost is very high. And secondly, identifying. The method comprises two modes, wherein in the first mode, a data set is manufactured according to the detection result of the detection algorithm in the first step, and a corresponding classifier is trained to classify the data set, so that the effect of the mode is limited by the performance of the detection algorithm; and secondly, manually selecting and segmenting the data set, training a classifier on the premise of ensuring that a training image contains a complete target to be identified, and segmenting the image by using a detection algorithm during testing, wherein the method is time-consuming and labor-consuming and increases the labor cost of the algorithm.
Disclosure of Invention
In view of the above, an object of the present invention is to overcome the defects in the prior art, and provide a dragonflies insect identification method based on a regional recommendation network, which can make identification processing simple and fast, save a large amount of labor cost, and solve the problem of difficult identification caused by a complex background of dragonflies insect pictures in a natural environment.
The invention relates to a dragonfly order insect identification method based on a regional suggestion network, which comprises the following steps of:
s1, acquiring an image of an insect of the dragonfly order, and cleaning and sorting the image to obtain a data set of the image of the insect of the dragonfly order;
s2, enhancing the data set of the dragonfly order insect image to obtain an enhanced data set;
s3, dividing the enhanced data set according to a set proportion to obtain a training set, a verification set and a test set of the dragonfly order insect images;
s4, constructing a deep convolutional network model based on the regional suggestion network;
s5, setting training parameters, and training the deep convolution network model by using the training set and the verification set to obtain a trained network model;
and S6, inputting the test set into the trained network model, and outputting to obtain a classification result of the test set.
Further, the dragonfly order insect image is a dragonfly order insect image in a natural environment.
Further, in step S2, the enhancing process of the data set of the image of the dragonflies insects specifically includes: performing random horizontal flipping on the data set and performing random center clipping on the data set.
Further, step S4 specifically includes:
s41, constructing a regional suggestion network;
s42, taking ResNet50 as a feature extraction network of the deep convolution network model, and taking the area suggestion network as a feature screening network of the deep convolution network model;
s43, determining a loss function of the deep convolutional network model.
Further, a loss function L of the deep convolutional network model is determined according to the following formula:
L＝L1+μL2；
wherein L is1Is a cross entropy loss function; l is2Is improved Focal local; mu is a set coefficient;
the above-mentioned
yiAs a true result of the current sample,
for selected sub-regionsPredicting the result;
the above-mentioned
Respectively taking alpha and gamma as control parameters, wherein alpha and gamma are the prediction probability of the current sample;
the above-mentioned
id is the sample number and classes is the total number of labels for the sample.
Further, in step S5, setting training parameters, and training the deep convolutional network model by using the training set and the verification set, specifically including:
s51, taking a pre-training weight model on the ImageNet data set as an initialization weight model of ResNet 50;
s52, setting input sizes of the training set and the verification set dragonfly order insect images and setting preset sizes of sub-regions;
and S53, using Batch Normalization to realize regularization, and optimizing a deep convolution network model by adopting a random gradient descent algorithm.
The invention has the beneficial effects that: according to the dragonfly order insect identification method based on the regional suggestion network, the ResNet50 is used as the feature extraction network, manual feature design and extraction are not needed, and the convolutional neural network is designed as the regional suggestion network for feature screening, so that the capability of extracting effective features from a complex background by a model is enhanced, and the problem of difficult identification caused by the complex background of dragonfly order insects in a natural environment is solved.
Drawings
The invention is further described below with reference to the following figures and examples:
FIG. 1 is a schematic flow diagram of the process of the present invention;
FIG. 2 is a diagram of the overall network framework of the method of the present invention;
FIG. 3 is a schematic diagram of a regional recommendation network in accordance with the present invention;
FIG. 4 is a schematic diagram of the depth separable convolution of the present invention;
FIG. 5 is a schematic diagram of pooling and reflooling of the present invention.
Detailed Description
The invention is further described with reference to the drawings, as shown in fig. 1:
the invention relates to a dragonfly order insect identification method based on a regional suggestion network, which comprises the following steps of:
s1, acquiring an image of an insect of the dragonfly order, and cleaning and sorting the image to obtain a data set of the image of the insect of the dragonfly order; the method for acquiring the image data of the dragonflies insects comprises various modes such as field shooting, laboratory shooting, network downloading and the like, so that the finally obtained data set of the dragonflies insects does not have the characteristic of equal quantity of various dragonflies insects under given conditions, and the whole data is distributed in a long tail shape.
S2, enhancing the data set of the dragonfly order insect image to obtain an enhanced data set;
s3, dividing the enhanced data set according to a set proportion to obtain a training set, a verification set and a test set of the dragonfly order insect images; wherein the set ratio is 4.5:4.5: 1.
S4, constructing a deep convolutional network model based on the regional suggestion network; the method comprises the steps of constructing a depth convolution network model based on a regional suggestion network, and omitting the step of detecting dragonflies insects firstly, so that a large amount of manual labeling work on data is avoided, and the depth convolution network model can directly identify dragonflies types in an image to be detected;
s5, setting training parameters, and training the deep convolution network model by using the training set and the verification set to obtain a trained network model;
and S6, inputting the test set into the trained network model, and outputting to obtain a classification result of the test set.
In this embodiment, the image of the insect of the order dragonfly is an image of the insect of the order dragonfly in a natural environment. Wherein, the collected images of the dragonflies insects comprise a very small amount of laboratory sample images.
In this embodiment, in step S2, the enhancing process performed on the data set of the image of the dragonflies insects specifically includes: performing random horizontal flipping on the data set and performing random center clipping on the data set.
In this embodiment, the step S4 specifically includes:
s41, constructing a regional suggestion network; wherein the regional proposed network is abbreviated as rpn (regional pro-social network), as shown in fig. 3, a yellow dotted block represents average pooling, a yellow borderless block represents maximum anti-pooling, and a green borderless block represents depth separable convolution; for an input picture with a size of H × W × C, after passing through the RPN, a set of H × W × C × k sub-regions with a preset number (k) and size (H × W) is obtained, and the k sub-regions are added to the original image to be input into the final classification network.
S42, taking ResNet50 as a feature extraction network of the deep convolution network model, and taking the area suggestion network as a feature screening network of the deep convolution network model; specifically, as shown in fig. 4 and 5, for the input picture F ∈ RH×W×CSelecting M sub-regions with different preset scales, performing 3 × 3 depth convolution plus 1 × 1 point convolution, sub-connecting 3 × 3 average pooling, then connecting 3 × 3 maximum inverse pooling (from which one branch is 1 × 1 convolution), then performing a group of depth separable convolutions (from which one branch is 1 × 1 convolution), then performing 3 × 3 standard convolution, then performing 1 × 1 convolution, using a ReLu function as an activation function in the sampling process, and finally cascading with two branches to obtain a feature map FRAnd then pre-classified. And according to the pre-classification result, selecting k regions with highest confidence level from the M regions as sub-regions for RPN screening as output S ═ R1,R2,…,Rk}。
S43, determining a loss function of the deep convolutional network model. Wherein the existence of the data set is solved by introducing the loss function and adding the super-parameter to control the ratio in the overall network lossThe problem of the imbalance of the number of various types of samples. Specifically, for an input picture F ∈ RH×W×CAnd output S ═ R1,R2,…,RkAnd after feature extraction is carried out by using ResNet50, adding the obtained feature maps and then classifying, wherein the classification loss is processed by adopting the loss function.
In this embodiment, the loss function L of the deep convolutional network model is determined according to the following formula:
L＝L1+μL2；
wherein L is1Is a cross entropy loss function; l is2Is improved Focal local; mu is a set coefficient;
the above-mentioned
yiAs a true result of the current sample,
a prediction result for the selected sub-region;
the above-mentioned
For the prediction probability of the current sample, α and γ are respectively control parameters, and the parameters are used to adjust the loss calculation weight of the easily classified sample in the data set, in this embodiment, the parameter α is set to 0.5, and the parameter γ is set to 2;
the above-mentioned
id is the sample number and classes is the total number of labels for the sample. In this embodiment, when the data set is prepared, the sample numbers (id) are set according to the descending order of the number of samples contained in each class, that is, the number of the first class is the largest, and the number of the last class is the smallest. When the label of the current training sample is in the first half of the total label number (classes) of the sample, μ ═ 0.5, otherwise μ ═ 1.
In this embodiment, in step S5, setting a training parameter, and training a deep convolutional network model using the training set and the verification set, specifically including:
s51, taking a pre-training weight model on the ImageNet data set as an initialization weight model of ResNet 50;
s52, setting input sizes of the training set and the verification set dragonfly order insect images and setting preset sizes of sub-regions;
and S53, using Batch Normalization to realize regularization, and optimizing a deep convolution network model by adopting a random gradient descent algorithm.
Specifically, in the training process, the Input size (Input size) of the image is set to 448 × 448 (pixel value), the preset sizes of the sub-regions are 48 × 48, 96 × 96, 192 × 192 (pixel value), k is 3, and bn (batch normalization) is used to implement regularization. The initial learning rate is 0.001 and the optimizer is Momentum SGD. NMS threshold 0.25, weight decay 0.0001, iteration epoch 200.
Finally, the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all of them should be covered in the claims of the present invention.
Claims (6)
1. A dragonfly order insect identification method based on a regional suggestion network is characterized in that: the method comprises the following steps:
s1, acquiring an image of an insect of the dragonfly order, and cleaning and sorting the image to obtain a data set of the image of the insect of the dragonfly order;
s2, enhancing the data set of the dragonfly order insect image to obtain an enhanced data set;
s3, dividing the enhanced data set according to a set proportion to obtain a training set, a verification set and a test set of the dragonfly order insect images;
s4, constructing a deep convolutional network model based on the regional suggestion network;
s5, setting training parameters, and training the deep convolution network model by using the training set and the verification set to obtain a trained network model;
and S6, inputting the test set into the trained network model, and outputting to obtain a classification result of the test set.
2. The method for identifying dragonflies insects based on a regional recommendation network as claimed in claim 1, wherein: the image of the dragonfly order insects is the image of the dragonfly order insects in the natural environment.
3. The method for identifying dragonflies insects based on a regional recommendation network as claimed in claim 1, wherein: in step S2, the enhancing process of the data set of the dragonflies insect image specifically includes: performing random horizontal flipping on the data set and performing random center clipping on the data set.
4. The method for identifying dragonflies insects based on a regional recommendation network as claimed in claim 1, wherein: the step S4 specifically includes:
s41, constructing a regional suggestion network;
s42, taking ResNet50 as a feature extraction network of the deep convolution network model, and taking the area suggestion network as a feature screening network of the deep convolution network model;
s43, determining a loss function of the deep convolutional network model.
5. The method for identifying dragonflies insects based on a regional recommendation network as claimed in claim 4, wherein: determining a loss function L of the deep convolutional network model according to the following formula:
L＝L1+μL2；
wherein L is1Is a cross entropy loss function; l is2Is improved Focal local;mu is a set coefficient;
the above-mentioned
yiAs a true result of the current sample,
a prediction result for the selected sub-region;
the above-mentioned
Respectively taking alpha and gamma as control parameters, wherein alpha and gamma are the prediction probability of the current sample;
the above-mentioned
id is the sample number and classes is the total number of labels for the sample.
6. The method for identifying dragonflies insects based on a regional recommendation network as claimed in claim 4, wherein: in step S5, setting training parameters, and training a deep convolutional network model using the training set and the validation set, specifically including:
s51, taking a pre-training weight model on the ImageNet data set as an initialization weight model of ResNet 50;
s52, setting input sizes of the training set and the verification set dragonfly order insect images and setting preset sizes of sub-regions;
and S53, using Batch Normalization to realize regularization, and optimizing a deep convolution network model by adopting a random gradient descent algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110480792.7A CN113177486B (en) | 2021-04-30 | 2021-04-30 | Dragonfly order insect identification method based on regional suggestion network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110480792.7A CN113177486B (en) | 2021-04-30 | 2021-04-30 | Dragonfly order insect identification method based on regional suggestion network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113177486A true CN113177486A (en) | 2021-07-27 |
| CN113177486B CN113177486B (en) | 2022-06-03 |
Family
ID=76925719
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110480792.7A Active CN113177486B (en) | 2021-04-30 | 2021-04-30 | Dragonfly order insect identification method based on regional suggestion network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113177486B (en) |
Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109359681A (en) * | 2018-10-11 | 2019-02-19 | 西京学院 | A kind of field crop pest and disease disasters recognition methods based on the full convolutional neural networks of improvement |
| US20190102646A1 (en) * | 2017-10-02 | 2019-04-04 | Xnor.ai Inc. | Image based object detection |
| CN111126385A (en) * | 2019-12-13 | 2020-05-08 | 哈尔滨工程大学 | Deep learning intelligent identification method for deformable living body small target |
| CN111444952A (en) * | 2020-03-24 | 2020-07-24 | 腾讯科技（深圳）有限公司 | Method and device for generating sample identification model, computer equipment and storage medium |
| CN111476302A (en) * | 2020-04-08 | 2020-07-31 | 北京工商大学 | fast-RCNN target object detection method based on deep reinforcement learning |
| CN111652247A (en) * | 2020-05-28 | 2020-09-11 | 大连海事大学 | Diptera insect identification method based on deep convolutional neural network |
| CN111898406A (en) * | 2020-06-05 | 2020-11-06 | 东南大学 | Face detection method based on focus loss and multitask cascade |
| CN111931581A (en) * | 2020-07-10 | 2020-11-13 | 威海精讯畅通电子科技有限公司 | Agricultural pest identification method based on convolutional neural network, terminal and readable storage medium |
| CN112070043A (en) * | 2020-09-15 | 2020-12-11 | 常熟理工学院 | Safety helmet wearing convolutional network based on feature fusion, training and detecting method |
| CN112116603A (en) * | 2020-09-14 | 2020-12-22 | 中国科学院大学宁波华美医院 | Pulmonary nodule false positive screening method based on multitask learning |
| WO2021008233A1 (en) * | 2019-07-17 | 2021-01-21 | 上海商汤智能科技有限公司 | Robot image enhancement method and apparatus, processor, device, medium and program |
| CN112257569A (en) * | 2020-10-21 | 2021-01-22 | 青海城市云大数据技术有限公司 | Target detection and identification method based on real-time video stream |
| CN112288795A (en) * | 2020-10-29 | 2021-01-29 | 深圳大学 | Insect density calculation method and device based on fast-RCNN |
| CN112464971A (en) * | 2020-04-09 | 2021-03-09 | 丰疆智能软件科技(南京)有限公司 | Method for constructing pest detection model |
| CN112465819A (en) * | 2020-12-18 | 2021-03-09 | 平安科技（深圳）有限公司 | Image abnormal area detection method and device, electronic equipment and storage medium |
| CN112598657A (en) * | 2020-12-28 | 2021-04-02 | 锋睿领创(珠海)科技有限公司 | Defect detection method and device, model construction method and computer equipment |
-
2021
- 2021-04-30 CN CN202110480792.7A patent/CN113177486B/en active Active
Patent Citations (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190102646A1 (en) * | 2017-10-02 | 2019-04-04 | Xnor.ai Inc. | Image based object detection |
| CN109359681A (en) * | 2018-10-11 | 2019-02-19 | 西京学院 | A kind of field crop pest and disease disasters recognition methods based on the full convolutional neural networks of improvement |
| WO2021008233A1 (en) * | 2019-07-17 | 2021-01-21 | 上海商汤智能科技有限公司 | Robot image enhancement method and apparatus, processor, device, medium and program |
| CN111126385A (en) * | 2019-12-13 | 2020-05-08 | 哈尔滨工程大学 | Deep learning intelligent identification method for deformable living body small target |
| CN111444952A (en) * | 2020-03-24 | 2020-07-24 | 腾讯科技（深圳）有限公司 | Method and device for generating sample identification model, computer equipment and storage medium |
| CN111476302A (en) * | 2020-04-08 | 2020-07-31 | 北京工商大学 | fast-RCNN target object detection method based on deep reinforcement learning |
| CN112464971A (en) * | 2020-04-09 | 2021-03-09 | 丰疆智能软件科技(南京)有限公司 | Method for constructing pest detection model |
| CN111652247A (en) * | 2020-05-28 | 2020-09-11 | 大连海事大学 | Diptera insect identification method based on deep convolutional neural network |
| CN111898406A (en) * | 2020-06-05 | 2020-11-06 | 东南大学 | Face detection method based on focus loss and multitask cascade |
| CN111931581A (en) * | 2020-07-10 | 2020-11-13 | 威海精讯畅通电子科技有限公司 | Agricultural pest identification method based on convolutional neural network, terminal and readable storage medium |
| CN112116603A (en) * | 2020-09-14 | 2020-12-22 | 中国科学院大学宁波华美医院 | Pulmonary nodule false positive screening method based on multitask learning |
| CN112070043A (en) * | 2020-09-15 | 2020-12-11 | 常熟理工学院 | Safety helmet wearing convolutional network based on feature fusion, training and detecting method |
| CN112257569A (en) * | 2020-10-21 | 2021-01-22 | 青海城市云大数据技术有限公司 | Target detection and identification method based on real-time video stream |
| CN112288795A (en) * | 2020-10-29 | 2021-01-29 | 深圳大学 | Insect density calculation method and device based on fast-RCNN |
| CN112465819A (en) * | 2020-12-18 | 2021-03-09 | 平安科技（深圳）有限公司 | Image abnormal area detection method and device, electronic equipment and storage medium |
| CN112598657A (en) * | 2020-12-28 | 2021-04-02 | 锋睿领创(珠海)科技有限公司 | Defect detection method and device, model construction method and computer equipment |
Non-Patent Citations (10)
| Title |
|---|
| CHEN Y等: ""Destruction and construction learning for fine-grained image recognition"", 《PROCEEDINGS OF THE 2019 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》 * |
| CHEN Y等: ""Destruction and construction learning for fine-grained image recognition"", 《PROCEEDINGS OF THE 2019 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION》, 31 December 2019 (2019-12-31), pages 5157 - 5166 * |
| YANG Z等: ""Learning to navigate for fine-grained classification"", 《PROCEEDINGS OF THE 2018 EUROPEAN CONFERENCE ON COMPUTER VISION》 * |
| YANG Z等: ""Learning to navigate for fine-grained classification"", 《PROCEEDINGS OF THE 2018 EUROPEAN CONFERENCE ON COMPUTER VISION》, 31 December 2018 (2018-12-31), pages 438 - 454 * |
| 李宇昕等: ""基于改进残差网络的道口车辆分类方法"", <《激光与光电子学进展》 * |
| 李宇昕等: ""基于改进残差网络的道口车辆分类方法"", <《激光与光电子学进展》, vol. 58, no. 4, 28 February 2021 (2021-02-28), pages 1 - 7 * |
| 翁雨辰等: ""深度区域网络方法的细粒度图像分类"", 《中国图象图形学报》 * |
| 翁雨辰等: ""深度区域网络方法的细粒度图像分类"", 《中国图象图形学报》, vol. 22, no. 11, 31 December 2017 (2017-12-31), pages 1521 - 1531 * |
| 赵浩如等: ""基于RPN与B-CNN的细粒度图像分类算法研究"", 《计算机应用与软件》 * |
| 赵浩如等: ""基于RPN与B-CNN的细粒度图像分类算法研究"", 《计算机应用与软件》, vol. 36, no. 3, 31 March 2019 (2019-03-31), pages 210 - 213 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113177486B (en) | 2022-06-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3478728B1 (en) | Method and system for cell annotation with adaptive incremental learning | |
| CN107609485B (en) | Traffic sign recognition method, storage medium and processing device | |
| CN109509187B (en) | Efficient inspection algorithm for small defects in large-resolution cloth images | |
| CN108898047B (en) | Pedestrian detection method and system based on blocking and shielding perception | |
| CN107316036B (en) | Insect pest identification method based on cascade classifier | |
| EP3690741B1 (en) | Method for automatically evaluating labeling reliability of training images for use in deep learning network to analyze images, and reliability-evaluating device using the same | |
| CN109460754B (en) | A kind of water surface foreign matter detecting method, device, equipment and storage medium | |
| CN111738064B (en) | Haze concentration identification method for haze image | |
| CN108564085B (en) | Method for automatically reading of pointer type instrument | |
| CN109242826B (en) | Mobile equipment end stick-shaped object root counting method and system based on target detection | |
| CN106951899A (en) | Method for detecting abnormality based on image recognition | |
| CN107871316B (en) | Automatic X-ray film hand bone interest area extraction method based on deep neural network | |
| EP3506165A1 (en) | Using a first stain to train a model to predict the region stained by a second stain | |
| CN111382766A (en) | Equipment fault detection method based on fast R-CNN | |
| CN105955708A (en) | Sports video lens classification method based on deep convolutional neural networks | |
| CN112926652B (en) | Fish fine granularity image recognition method based on deep learning | |
| CN109919149B (en) | Object labeling method and related equipment based on object detection model | |
| CN113435407B (en) | Small target identification method and device for power transmission system | |
| CN105913090B (en) | SAR image objective classification method based on SDAE-SVM | |
| CN111738114B (en) | Vehicle target detection method based on anchor-free accurate sampling remote sensing image | |
| CN110059539A (en) | A kind of natural scene text position detection method based on image segmentation | |
| CN111414951B (en) | Fine classification method and device for images | |
| CN111340019A (en) | Grain bin pest detection method based on Faster R-CNN | |
| CN111611889A (en) | Miniature insect pest recognition device in farmland based on improved convolutional neural network | |
| CN111814820B (en) | Image processing method and device |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |