CN109409226B - Finger vein image quality evaluation method and device based on cascade optimization CNN - Google Patents
Finger vein image quality evaluation method and device based on cascade optimization CNN Download PDFInfo
- Publication number
- CN109409226B CN109409226B CN201811113400.8A CN201811113400A CN109409226B CN 109409226 B CN109409226 B CN 109409226B CN 201811113400 A CN201811113400 A CN 201811113400A CN 109409226 B CN109409226 B CN 109409226B
- Authority
- CN
- China
- Prior art keywords
- training
- cnn
- quality
- cascade optimization
- finger vein
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- 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
- G06F18/2148—Generating training patterns; Bootstrap methods, e.g. bagging or boosting characterised by the process organisation or structure, e.g. boosting cascade
-
- 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
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/14—Vascular patterns
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a finger vein image quality evaluation method and device based on cascade optimization CNN. Sending the gray level image to be tested of the finger vein image to a pre-trained cascade optimization CNN for feature extraction, classifying according to the extracted features and a preset quality candidate class, and finishing the quality evaluation of the finger vein image by taking a classification result as an evaluation result. Simplifying the final model and realizing high-efficiency and low-cost quality evaluation.
Description
Technical Field
The invention relates to the field of biological identification, in particular to a finger vein image quality evaluation method and device based on cascade optimization CNN.
Background
At present, finger vein recognition is increasingly popularized in the application process as a new biological feature recognition technology. Due to the difference between users and devices, the quality of the finger vein map acquired each time is also high or low, the low-quality finger vein map is generally fuzzy, if the low-quality finger vein map is registered in the database, feature extraction and matching during later application are seriously influenced, and the identification performance of the system is poor, so that quality evaluation is required after each acquisition, and the high-quality finger vein map is ensured to be registered in the database. In the prior art, a CNN (convolutional neural network algorithm) is mostly adopted to convert an acquired finger vein image into a gray image and a binary image, then feature extraction is performed, and the extracted features are superposed for evaluation.
Disclosure of Invention
In order to solve the above problems, the present invention aims to provide a method that can directly send a gray scale image to be measured to the cascade optimization CNN pre-trained by a binary image to complete quality evaluation, and only a single pre-trained model is introduced in practical application, so that model space is saved, and the quality evaluation of a finger vein image more easily and end-to-end is realized.
The technical scheme adopted by the invention for solving the problems is as follows: a finger vein image quality evaluation method based on cascade optimization CNN comprises the following steps:
reading an input gray-scale image to be tested, and sending the gray-scale image to be tested to a pre-trained cascade optimization CNN;
extracting the characteristics of the gray level graph to be detected in the cascade optimization CNN to obtain a characteristic vector;
reading a preset quality candidate class, performing flexible maximum calculation on the characteristic vectors, classifying the characteristic vectors into corresponding quality candidate classes, and finishing evaluation.
Further, the training method for the pre-trained cascade optimization CNN includes:
reading a training binary image, and sending the training binary image to a pre-training CNN for learning to obtain a primary evaluation model;
and reading the training gray level image and sending the training gray level image to a primary evaluation model, wherein the primary evaluation model learns the characteristics of the training gray level image to obtain the cascade optimization CNN.
Further, the pre-training CNN includes 4 layers of convolutional networks and 2 layers of fully-connected layers, where the convolutional networks include convolutional layers, pooling layers, and random deactivation layers.
Further, the convolution kernel of the convolution layer is 3 × 3, and the convolution step size is 1; the template of the pooling layer is 2 multiplied by 2, and the step length is 2.
Further, the fully-connected layer includes a first fully-connected layer including 512 neurons and a second fully-connected layer including 2 neurons.
Further, the quality candidate classes include a high quality class and a low quality class.
A finger vein map quality evaluation device based on cascade optimization CNN comprises the following devices:
the device for acquiring the gray level image to be tested is used for reading the input gray level image to be tested and sending the gray level image to be tested to the pre-trained cascade optimization CNN;
a feature vector obtaining device, configured to perform feature extraction on the to-be-detected grayscale map in the cascade optimization CNN to obtain a feature vector;
and the quality evaluation device is used for reading a preset quality candidate class, performing flexible maximum calculation on the characteristic vector, classifying the characteristic vector into a corresponding quality candidate class and finishing evaluation.
Further, the training of the pre-trained cascade optimization CNN further includes the following apparatus:
the primary evaluation model acquisition device is used for reading a training binary image, sending the training binary image to a pre-training CNN for learning, and acquiring a primary evaluation model;
and the cascade optimization CNN acquisition device is used for reading the training gray level graph and sending the training gray level graph to the primary evaluation model, and the primary evaluation model learns the characteristics of the training gray level graph to obtain the cascade optimization CNN.
A finger vein map quality assessment device based on cascade optimization CNN comprises:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a cascaded optimization CNN based finger vein graph quality assessment method as described above.
A non-transitory computer-readable storage medium storing computer-executable instructions for causing a computer to perform a cascade optimization CNN-based finger vein map quality assessment method as described above.
The invention has the beneficial effects that: the invention adopts a finger vein image quality evaluation method and a device thereof based on cascade optimization CNN. And sending the input gray level image to be detected to the pre-trained cascade optimization CNN for feature extraction, and calculating the quality classification according to the extracted feature vector, thereby realizing the quality evaluation of the finger vein image. Compared with the prior art, the method for simultaneously training the binary image and the gray image and extracting the features has the advantages that one feature extraction network is reduced by introducing the pre-trained cascade optimization CNN, so that the model for extraction is simpler and more convenient, the recognition speed is increased, and a large amount of storage space and cost are saved.
Drawings
The invention is further illustrated with reference to the following figures and examples.
FIG. 1 is a flowchart of a method for evaluating quality of a finger vein map based on cascade optimization CNN according to the present invention;
FIG. 2 is a flowchart of a cascaded optimization CNN training process of the finger vein image quality evaluation method based on cascaded optimization CNN of the present invention;
FIG. 3 is a schematic diagram of cascaded optimization CNN training of the finger vein image quality evaluation method based on cascaded optimization CNN of the present invention;
fig. 4 is a schematic diagram of a convolutional network structure of a pre-trained CNN according to a first embodiment of the finger vein diagram quality assessment method based on cascade optimization CNN of the present invention;
FIG. 5 is a detailed step diagram of the finger vein image quality assessment method based on cascade optimization CNN according to the present invention;
fig. 6 is a schematic diagram of a convolutional network structure of a pre-trained CNN according to a second embodiment of the finger vein diagram quality assessment method based on cascade optimization CNN of the present invention;
fig. 7 is a schematic diagram of a convolutional network structure of a pre-trained CNN according to a third embodiment of the finger vein diagram quality assessment method based on cascade optimization CNN of the present invention.
The reference numbers illustrate:
1. training a binary image; 2. training a gray scale map; 3. pre-training a CNN; 4. a fully-connected layer; 5. a convolutional network structure.
Detailed Description
Referring to fig. 1, the method for evaluating the quality of a finger vein map based on cascade optimization CNN of the present invention includes the following steps:
reading an input gray-scale image to be tested, and sending the gray-scale image to be tested to a pre-trained cascade optimization CNN;
extracting the characteristics of the gray level graph to be detected in the cascade optimization CNN to obtain a characteristic vector;
reading a preset quality candidate class, performing flexible maximum calculation on the characteristic vectors, classifying the characteristic vectors into corresponding quality candidate classes, and finishing evaluation.
The method is characterized in that a gray-scale image is taken as a main figure collected by most common identification equipment, and a binary operation method is also adopted for obtaining the binary image, so that the equipment can be simplified by directly extracting the gray-scale image to be detected.
After the feature vector extracted in the cascade optimization CNN is subjected to flexible maximum calculation, the probability of each candidate class in the quality candidate class corresponding to the feature vector can be obtained, and the class with the maximum probability is judged as the class during evaluation, so that an evaluation result is obtained.
For example, if the calculated feature vector has a high-quality class probability of 70% and a low-quality class probability of 30%, the input image is evaluated as a high-quality image.
With reference to fig. 2 to fig. 3, further, the training method for the pre-trained cascade optimization CNN includes:
reading a training binary image 1, and sending the training binary image 1 to a pre-training CNN3 for learning to obtain a primary evaluation model;
and reading the training gray level image 2 and sending the training gray level image to a primary evaluation model, wherein the primary evaluation model learns the characteristics of the training gray level image to obtain the cascade optimization CNN.
When the pre-training CNN3 is used for learning the training binary image 1 and the primary evaluation model is used for training the training gray image 2, the selected deep learning frame is a keras module in Tensorflow, the optimizer is an ADAM optimizer, the initial learning rate is set to be 0.0003, and the learning times are set to be 1000 times.
Preferably, the pre-trained CNN or the primary evaluation model reads the initial learning rate of 0.0003 after the learning starts, and reads the initial LOSS value at the same time, and when the number of times of learning reaches 200 times, reads the current LOSS value again, and if the current LOSS value is equal to the initial LOSS value, it is determined that the learning network has converged, that is, the learning is completed. If the current LOSS value is smaller than the initial LOSS value, the learning rate is changed to be 0.1 time of the initial original rate, training is carried out for 200 times, and iteration is carried out in sequence until the LOSS value is not changed or the training is finished for 1000 times.
Referring to fig. 4, further, the pre-training CNN includes 4 layers of convolutional networks and 2 layers of fully-connected layers, where the convolutional networks include convolutional layers, pooling layers, and random deactivation layers.
Further, the convolution kernel of the convolution layer is 3 × 3, and the convolution step size is 1; the template of the pooling layer is 2 multiplied by 2, and the step length is 2.
Further, the fully-connected layer includes a first fully-connected layer including 512 neurons and a second fully-connected layer including 2 neurons.
The feature vector calculated by the convolutional layer is a 512-dimensional vector, so that the 512-dimensional vector can be converted into a 2-dimensional feature vector after passing through the first fully-connected layer and the second fully-connected layer, and the classification in the subsequent steps is facilitated.
Further, the quality candidate classes include a high quality class and a low quality class.
Preferably, the high-quality class and the low-quality class of the quality candidate classes include pre-labeled classification features, that is, the high-quality class includes features corresponding to the high-quality image, and the low-quality class includes features corresponding to the low-quality image. When the classification is carried out, probability vectors which are consistent between each feature in the feature vectors and features in the quality candidate class are calculated by utilizing the flexibility maximum, and the probability vectors of all the features belonging to the same quality candidate class are added to obtain the probability vector corresponding to the quality candidate class.
Referring to fig. 5, the following is an illustration of the overall process steps of the present invention:
step 101, reading an input gray-scale image to be detected, and sending the gray-scale image to be detected to a pre-trained cascade optimization CNN;
102, performing feature extraction on the gray level image through cascade optimization CNN to obtain 512-dimensional feature vectors;
103, reading a preset quality candidate class, sending the characteristic vector to a full-connection layer, and performing flexible maximum calculation in the full-connection layer to obtain a classification probability vector;
step 104, if the classification probability vector is a low-quality classification, setting the input gray level image as a low-quality finger vein image; and if the classification probability vector is high-quality classification, setting the input gray level image as a high-quality finger vein image.
A finger vein map quality evaluation device based on cascade optimization CNN comprises the following devices:
the device for acquiring the gray level image to be tested is used for reading the input gray level image to be tested and sending the gray level image to be tested to the pre-trained cascade optimization CNN;
a feature vector obtaining device, configured to perform feature extraction on the to-be-detected grayscale map in the cascade optimization CNN to obtain a feature vector;
and the quality evaluation device is used for reading a preset quality candidate class, performing flexible maximum calculation on the characteristic vector, classifying the characteristic vector into a corresponding quality candidate class and finishing evaluation.
Further, the training of the pre-trained cascade optimization CNN further includes the following apparatus:
the primary evaluation model acquisition device is used for reading a training binary image, sending the training binary image to a pre-training CNN for learning, and acquiring a primary evaluation model;
and the cascade optimization CNN acquisition device is used for reading the training gray level graph and sending the training gray level graph to the primary evaluation model, and the primary evaluation model learns the characteristics of the training gray level graph to obtain the cascade optimization CNN.
A finger vein map quality assessment device based on cascade optimization CNN comprises:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform a cascaded optimization CNN based finger vein graph quality assessment method as described above.
A non-transitory computer-readable storage medium storing computer-executable instructions for causing a computer to perform a cascade optimization CNN-based finger vein map quality assessment method as described above.
Referring to fig. 6, a method for evaluating quality of a finger vein map based on cascade optimization CNN has a basic flow substantially the same as that of the first embodiment, with the following differences: the pre-training CNN comprises a convolutional network with 3 layers, wherein the activation function of the first 2 layers adopts a Relu activation function, and the layer 3 adopts a softmax activation function; the random deactivation probability of the random deactivation layer of the first 2 layers of convolutional networks is 0.3, and the random deactivation probability of the random deactivation layer of the 3 rd layer of convolutional networks is 0.5.
Referring to fig. 7, a method for evaluating quality of a finger vein map based on cascade optimization CNN has a basic flow substantially the same as that of the first embodiment, with the following differences: the pre-training CNN comprises a convolutional network with 5 layers, wherein the activation function of the first 4 layers adopts a Relu activation function, and the layer 5 adopts a softmax activation function; the random deactivation probability of the random deactivation layer of the first 2 layers of convolutional networks is 0.3, and the random deactivation probability of the random deactivation layer of the last 3 layers of convolutional networks is 0.5.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a general hardware platform, and certainly can also be implemented by hardware. It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a computer readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. The storage medium may be a magnetic disk, an optical disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), or the like.
The above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiment, and the present invention shall fall within the protection scope of the present invention as long as the technical effects of the present invention are achieved by the same means.
Claims (8)
1. A finger vein image quality evaluation method based on cascade optimization CNN is characterized by comprising the following steps:
reading an input gray-scale image to be tested, and sending the gray-scale image to be tested to a pre-trained cascade optimization CNN;
extracting the characteristics of the gray level graph to be detected in the cascade optimization CNN to obtain a characteristic vector;
reading a preset quality candidate class, performing flexible maximum calculation on the characteristic vector, classifying the characteristic vector into a corresponding quality candidate class, and finishing evaluation;
the training method of the pre-trained cascade optimization CNN comprises the following steps:
reading a training binary image, and sending the training binary image to a pre-training CNN for learning to obtain a primary evaluation model;
and reading the training gray level image and sending the training gray level image to a primary evaluation model, wherein the primary evaluation model learns the characteristics of the training gray level image to obtain the cascade optimization CNN.
2. The finger vein map quality assessment method based on cascade optimization CNN as claimed in claim 1, characterized in that: the pre-training CNN comprises 4 layers of convolutional networks and 2 layers of full-connection layers, wherein the convolutional networks comprise convolutional layers, pooling layers and random deactivation layers.
3. The finger vein map quality assessment method based on cascade optimization CNN as claimed in claim 2, characterized in that: the convolution kernel of the convolution layer is 3 multiplied by 3, and the convolution step length is 1; the template of the pooling layer is 2 multiplied by 2, and the step length is 2.
4. The finger vein map quality assessment method based on cascade optimization CNN as claimed in claim 2, characterized in that: the fully-connected layer comprises a first fully-connected layer and a second fully-connected layer, the first fully-connected layer comprises 512 neurons, and the second fully-connected layer comprises 2 neurons.
5. The method for evaluating the quality of the finger vein map based on the cascade optimization CNN as claimed in claim 1, wherein the quality candidate classes comprise a high quality class and a low quality class.
6. A finger vein map quality evaluation device based on cascade optimization CNN is characterized by comprising the following devices:
the device for acquiring the gray level image to be tested is used for reading the input gray level image to be tested and sending the gray level image to be tested to the pre-trained cascade optimization CNN;
a feature vector obtaining device, configured to perform feature extraction on the to-be-detected grayscale map in the cascade optimization CNN to obtain a feature vector;
the quality evaluation device is used for reading a preset quality candidate class, performing flexible maximum calculation on the characteristic vector, classifying the characteristic vector into a corresponding quality candidate class and finishing evaluation;
the training of the pre-trained cascade optimization CNN further comprises the following means:
the primary evaluation model acquisition device is used for reading a training binary image, sending the training binary image to a pre-training CNN for learning, and acquiring a primary evaluation model;
and the cascade optimization CNN acquisition device is used for reading the training gray level graph and sending the training gray level graph to the primary evaluation model, and the primary evaluation model learns the characteristics of the training gray level graph to obtain the cascade optimization CNN.
7. A finger vein map quality assessment device based on cascade optimization CNN is characterized by comprising:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor;
wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method of any one of claims 1-5.
8. A non-transitory computer-readable storage medium, characterized in that: the computer-readable storage medium stores computer-executable instructions for causing a computer to perform the method of any one of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811113400.8A CN109409226B (en) | 2018-09-25 | 2018-09-25 | Finger vein image quality evaluation method and device based on cascade optimization CNN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811113400.8A CN109409226B (en) | 2018-09-25 | 2018-09-25 | Finger vein image quality evaluation method and device based on cascade optimization CNN |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109409226A CN109409226A (en) | 2019-03-01 |
| CN109409226B true CN109409226B (en) | 2022-04-08 |
Family
ID=65465062
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811113400.8A Active CN109409226B (en) | 2018-09-25 | 2018-09-25 | Finger vein image quality evaluation method and device based on cascade optimization CNN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109409226B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110738141A (en) * | 2019-09-26 | 2020-01-31 | 五邑大学 | vein identification method, device, equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102915528A (en) * | 2012-11-06 | 2013-02-06 | 杭州电子科技大学 | Method for enhancing binary image of array cascade FHN (FitzHugh Nagumo) model stochastic resonance mechanism |
| CN106529468A (en) * | 2016-11-07 | 2017-03-22 | 重庆工商大学 | Finger vein identification method and system based on convolutional neural network |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8872909B2 (en) * | 2010-06-10 | 2014-10-28 | The Hong Kong Polytechnic University | Method and apparatus for personal identification using finger imaging |
| CN101847208B (en) * | 2010-06-11 | 2012-06-06 | 哈尔滨工程大学 | Secondary classification fusion identification method for fingerprint and finger vein bimodal identification |
| CN106326886B (en) * | 2016-11-07 | 2019-05-10 | 重庆工商大学 | Finger vein image quality appraisal procedure based on convolutional neural networks |
-
2018
- 2018-09-25 CN CN201811113400.8A patent/CN109409226B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102915528A (en) * | 2012-11-06 | 2013-02-06 | 杭州电子科技大学 | Method for enhancing binary image of array cascade FHN (FitzHugh Nagumo) model stochastic resonance mechanism |
| CN106529468A (en) * | 2016-11-07 | 2017-03-22 | 重庆工商大学 | Finger vein identification method and system based on convolutional neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109409226A (en) | 2019-03-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR102641116B1 (en) | Method and device to recognize image and method and device to train recognition model based on data augmentation | |
| CN109934197B (en) | Training method and device for face recognition model and computer readable storage medium | |
| CN109840531B (en) | Method and device for training multi-label classification model | |
| CN110633745B (en) | Image classification training method and device based on artificial intelligence and storage medium | |
| CN109410974B (en) | Voice enhancement method, device, equipment and storage medium | |
| CN109002755B (en) | Age estimation model construction method and estimation method based on face image | |
| CN109325440B (en) | Human body action recognition method and system | |
| CN110135505B (en) | Image classification method and device, computer equipment and computer readable storage medium | |
| CN113034483B (en) | Cigarette defect detection method based on deep migration learning | |
| CN110288085B (en) | Data processing method, device and system and storage medium | |
| CN110837830B (en) | Image character recognition method based on space-time convolutional neural network | |
| CN112381763A (en) | Surface defect detection method | |
| CN115035418A (en) | Remote sensing image semantic segmentation method and system based on improved deep LabV3+ network | |
| CN112364974B (en) | YOLOv3 algorithm based on activation function improvement | |
| CN112861718A (en) | Lightweight feature fusion crowd counting method and system | |
| CN111695640A (en) | Foundation cloud picture recognition model training method and foundation cloud picture recognition method | |
| CN112418175A (en) | Rolling bearing fault diagnosis method and system based on domain migration and storage medium | |
| CN114283350A (en) | Visual model training and video processing method, device, equipment and storage medium | |
| CN111079837A (en) | Method for detecting, identifying and classifying two-dimensional gray level images | |
| CN109978058B (en) | Method, device, terminal and storage medium for determining image classification | |
| CN106384587A (en) | Voice recognition method and system thereof | |
| CN109409226B (en) | Finger vein image quality evaluation method and device based on cascade optimization CNN | |
| CN109101984B (en) | Image identification method and device based on convolutional neural network | |
| CN109101858B (en) | Action recognition method and device | |
| CN113283388A (en) | Training method, device and equipment of living human face detection model and storage medium |
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 |