CN116525145A - Facial dynamic video image pain assessment method based on dynamic fusion module - Google Patents
Facial dynamic video image pain assessment method based on dynamic fusion module Download PDFInfo
- Publication number
- CN116525145A CN116525145A CN202310526165.1A CN202310526165A CN116525145A CN 116525145 A CN116525145 A CN 116525145A CN 202310526165 A CN202310526165 A CN 202310526165A CN 116525145 A CN116525145 A CN 116525145A
- Authority
- CN
- China
- Prior art keywords
- pain
- model
- fusion module
- dynamic
- training
- 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
- 208000002193 Pain Diseases 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 title claims abstract description 36
- 230000004927 fusion Effects 0.000 title claims abstract description 29
- 230000001815 facial effect Effects 0.000 title claims description 20
- 238000012549 training Methods 0.000 claims abstract description 31
- 238000011156 evaluation Methods 0.000 claims abstract description 15
- 238000000605 extraction Methods 0.000 claims abstract description 12
- 206010028391 Musculoskeletal Pain Diseases 0.000 claims abstract description 11
- 208000007613 Shoulder Pain Diseases 0.000 claims abstract description 11
- 206010016059 Facial pain Diseases 0.000 claims abstract description 7
- 238000012545 processing Methods 0.000 claims abstract description 7
- 230000006872 improvement Effects 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims description 14
- 238000002790 cross-validation Methods 0.000 claims description 4
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 2
- 230000003068 static effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 230000002996 emotional effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000004445 quantitative analysis Methods 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000012800 visualization Methods 0.000 description 1
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/48—Other medical applications
- A61B5/4824—Touch or pain perception evaluation
-
- 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
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- 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
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- 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
-
- 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
- G06V40/16—Human faces, e.g. facial parts, sketches or expressions
- G06V40/174—Facial expression recognition
- G06V40/176—Dynamic expression
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H80/00—ICT specially adapted for facilitating communication between medical practitioners or patients, e.g. for collaborative diagnosis, therapy or health monitoring
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Multimedia (AREA)
- Pathology (AREA)
- Mathematical Physics (AREA)
- Public Health (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Databases & Information Systems (AREA)
- Primary Health Care (AREA)
- Heart & Thoracic Surgery (AREA)
- Human Computer Interaction (AREA)
- Psychiatry (AREA)
- Pain & Pain Management (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Hospice & Palliative Care (AREA)
- Animal Behavior & Ethology (AREA)
- Epidemiology (AREA)
- Veterinary Medicine (AREA)
- Surgery (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a face dynamic video image pain assessment method based on a dynamic fusion module, which comprises the steps of firstly carrying out data balance processing on a UNBC-McMaster Shoulder Pain Expression Archive data set, then in a pre-training model Resnet101 network, obtaining the dynamic fusion module based on Non-local framework improvement, establishing an image space feature extraction model, extracting the space feature information of each frame of image through the space feature extraction model, inputting the extracted feature information into a long-short-time memory network to extract the corresponding time sequence information of a face dynamic video image, further extracting the time sequence information between adjacent frames, realizing the identification of face pain, and finally realizing the automatic assessment of pain. The method aims at assisting doctors in carrying out long-time pain monitoring and evaluation on patients, provides an auxiliary means for clinical pain evaluation, and has potential application prospects for evaluating pain scenes of special clinical patients.
Description
Technical Field
The invention belongs to the technical field of computer vision, and particularly relates to a facial dynamic video image pain assessment method based on a dynamic fusion module.
Background
Pain is a complex phenomenon involving the sensory and emotional experience of the human body, and is also a self-protecting mechanism of the human body. In clinical settings, pain intensity estimation is critical to diagnosis and treatment of patients, and particularly has broad clinical application prospects for patients who cannot express pain in language. Self-reporting and observation methods are commonly used clinically as the mainstream methods of pain assessment, and these two methods are highly subjective and are not suitable for patients of a particular population. Related studies have shown that human faces contain abundant pain features, which makes automatic assessment of pain by extracting facial features an effective way.
To date, most of pain assessment studies based on facial pain expression are based on static image development-related studies, and the lack of time sequence information between adjacent frames in the feature extraction process of a static image may result in poor assessment performance. In addition, in a real clinical environment, pain of a patient needs to be monitored for a long time, which cannot be satisfied by a pain assessment method based on a static facial image. The pain evaluation based on the facial dynamic video image can meet the clinical requirements both on an offline basis and in a real-time basis.
Disclosure of Invention
In order to solve the technical problems, the invention provides a facial dynamic video image pain assessment method based on a dynamic fusion module.
The invention adopts the technical scheme that: a facial dynamic video image pain assessment method based on a dynamic fusion module comprises the following specific steps:
s1, carrying out data balance processing on a UNBC-McMaster Shoulder Pain Expression Archive data set;
s2, in a pre-training model Resnet101 network, a dynamic fusion module is obtained based on Non-local framework improvement, and an image space feature extraction model is established;
s3, extracting spatial feature information of each frame of image based on the spatial feature extraction model in the step S2, inputting the extracted feature information into a long-short-time memory network to extract corresponding time sequence information of the facial dynamic video image, further extracting the time sequence information between adjacent frames, realizing recognition of facial pain, and finally realizing automatic evaluation of the pain.
Further, the step S1 specifically includes the following steps:
the UNBC-McMaster Shoulder Pain Expression Archive dataset contained 25 subjects, 200 videos, 48398 frames of images, each of which was labeled for pain level using the PSPI method.
The data balancing processing mode specifically comprises the following steps:
(1) Deleting the complete sequence containing the pain-free frames;
(2) If the beginning or ending frame of a complete sequence contains a pain-free frame, then the portion is deleted.
Further, in the step S2, the dynamic fusion module is configured to fuse features extracted by a fourth block and a fifth block of the pretrained model Resnet101 network, which is specifically as follows:
features extracted by a fourth block and a fifth block of the pre-training model Resnet101 network are respectively marked as X h 、X l I.e. a high-level feature mapAnd a low-level feature map +.>
Wherein N is h 、N l Respectively represent X h 、X l Spatial position number, C h 、C l Respectively represent X h 、X l Is used for the number of channels of a computer,representing a feature set.
Fusing all extracted features of a fourth block and a fifth block in the Resnet101 network, inputting the fused features into a multi-scale channel attention module to calculate a weight value alpha for weighting X h And 1-alpha is used to weight X l Is a feature of the input of (a); the obtained characteristics weighted by the weight value alpha and 1-alpha are respectively used for X' h 、X l 'A'; and after the dynamic fusion module, extracting the spatial characteristic information of each input image frame.
Further, in the step S2, the pre-training model divides the data set by using a five-fold cross-validation method, so as to train and test the proposed network model, which is specifically as follows:
1) Training and testing the ratio of the data sets to be 4:1, training by using an Adam algorithm, updating parameters in a network model and storing the model;
2) And (3) repeating the step (1) for iterative training, and when the training is finished and the evaluation index reaches the highest, the model is called as an optimal model, and the model is stored.
The invention has the beneficial effects that: the method comprises the steps of carrying out data balance processing on UNBC-McMaster Shoulder Pain Expression Archive data sets, then in a pre-training model Resnet101 network, improving based on a Non-local framework to obtain a dynamic fusion module, establishing an image space feature extraction model, extracting space feature information of each frame of image through the space feature extraction model, inputting the extracted feature information into a long-short-term memory network to extract corresponding time sequence information of a face dynamic video image, further extracting time sequence information between adjacent frames, realizing recognition of facial pain, and finally realizing automatic evaluation of pain. The method aims at assisting doctors in carrying out long-time pain monitoring and evaluation on patients, provides an auxiliary means for clinical pain evaluation, and has potential application prospects for evaluating pain scenes of special clinical patients.
Drawings
Fig. 1 is a flowchart of a facial dynamic video image pain assessment method based on a dynamic fusion module according to the present invention.
FIG. 2 is a diagram showing a frame number distribution of each of the tested pain/no-pain videos provided by the UNBC-McMaster Shoulder Pain Expression Archive data set in accordance with an embodiment of the present invention.
FIG. 3 is a graph showing the frame number distribution of each of the tested pain/nopain videos in the balanced UNBC-McMaster dataset in accordance with an embodiment of the present invention.
Fig. 4 is a diagram of an overall network model in an embodiment of the invention.
FIG. 5 is a graph showing the average performance comparison result of different evaluation indexes of different input video frames in the Resnet101-LSTM model in the embodiment of the invention.
Detailed Description
The invention is further described below with reference to the drawings and examples.
As shown in fig. 1, the method for evaluating facial dynamic video image pain based on the dynamic fusion module comprises the following specific steps:
s1, carrying out data balance processing on a UNBC-McMaster Shoulder Pain Expression Archive data set;
s2, in a pre-training model Resnet101 network, a dynamic fusion module is obtained based on Non-local framework improvement, and an image space feature extraction model is established;
s3, extracting spatial feature information of each frame of image based on the spatial feature extraction model in the step S2, inputting the extracted feature information into a long-short-time memory network to extract corresponding time sequence information of the facial dynamic video image, further extracting the time sequence information between adjacent frames, realizing recognition of facial pain, and finally realizing automatic evaluation of the pain.
In this embodiment, the step S1 is specifically as follows:
the UNBC-McMaster Shoulder Pain Expression Archive dataset contained 25 subjects, 200 videos, 48398 frames of images, each of which was labeled for pain level using the PSPI method. As shown in fig. 2, it can be seen that the number of videos, the number of video frames, and the number of pain frames and the number of non-pain frames included in each test are unbalanced.
Where the number of non-painful frames is far greater than the number of painful frames, any model will tend to predict non-painful frames when pain recognition is performed based on the data distribution characteristics of the data set. The data and the corresponding balancing process are required to be performed, and the data balancing process is specifically performed as follows:
(1) Deleting the complete sequence containing the pain-free frames;
(2) If the beginning or ending frame of a complete sequence contains a pain-free frame, then the portion is deleted.
The painless image frames were removed by the method described above, and after the data set was balanced, each test contained the pain and non-pain frames as shown in fig. 3. Secondly, two methods of leave-one-out method and five-fold cross validation are adopted for the balanced data set to train the model provided by the invention.
In this embodiment, in the step S2, the dynamic fusion module is configured to fuse features extracted by a fourth block and a fifth block of the pretrained model Resnet101 network, which is specifically as follows:
in a pre-training model Resnet101 network, a dynamic fusion module is provided, the module takes Non-local as a main body frame, but because the existing Non-local frame has the defect of high calculation cost, an asymmetric fusion Non-local block module which is improved on the basis of Non-local is taken as an integral frame of the dynamic fusion module. The proposed dynamic fusion module is mainly used for fusing all extracted features of the fourth block and the fifth block of the Resnet101 network.
Features extracted by a fourth block and a fifth block of the pre-training model Resnet101 network are respectively marked as X h 、X l I.e. a high-level feature mapAnd a low-level feature map +.>
Wherein N is h 、N l Respectively represent X h 、X l Spatial position number, C h 、C l Respectively represent X h 、X l Is used for the number of channels of a computer,representing a feature set.
It is contemplated that features extracted by different blocks of the Resnet101 network may contribute differently to the final recognition accuracy. Fusing all extracted features of a fourth block and a fifth block in the Resnet101 network, inputting the fused features into a multi-scale channel attention module to calculate a weight value alpha for weighting X h And 1-alpha is used to weight X l Is a function of the input features of the computer.
The obtained characteristics after weighting by the weight value alpha and 1-alpha are respectively X h ′、X l 'A'; it should be noted that the multi-scale channel attention module is shown in FIG. 4 as MS-CAM, and that the MS-CAM uses X to represent input features; and Y represents an output feature; C. h, W the number of channels, height and width of the feature map, respectively; r represents a variable parameter, r=2 being set in the present embodiment; and after the dynamic fusion module, extracting the spatial characteristic information of each input image frame.
In this embodiment, in step S2, the pre-training model divides the data set by using a five-fold cross-validation method, so as to train and test the proposed network model, which is specifically as follows:
1) Training and testing the ratio of the data sets to be 4:1, training by using an Adam algorithm, updating parameters in a network model and storing the model.
2) And (3) repeating the step (1) for iterative training, and when the training is finished and the evaluation index reaches the highest, the model is called as an optimal model, and the model is stored.
As shown in fig. 4, the pre-training model Resnet101 network model extracts spatial feature information of each frame of image by using the pre-training model Resnet101, then inputs the extracted spatial feature information into a long-short-term memory network to extract corresponding time sequence information of a facial dynamic video image, and finally realizes recognition of facial pain.
In this embodiment, the image space feature extraction model provided by the present invention is also tested and evaluated, and specifically includes the following steps:
firstly, importing optimal model parameters obtained by training a pre-training model in the step S2, and inputting a test set image to test the performance of the model;
in the embodiment, the situation that the number of the video frames input by the proposed model is too small, the time sequence information is less, and enough time sequence information is not extracted is considered; and excessive input video frames can lead to redundancy of time sequence information. Therefore, the number of video frames input by selecting the model has an important meaning for pain identification accuracy.
In this example, the frequency frames were selected from 10 frames, 15 frames, 20 frames, 25 frames, 30 frames, 35 frames and 40 frames for training and testing the baseline model (Resnet 101-LSTM model), and the obtained baseline results are shown in Table 1. Table 1 shows the results of quantitative analysis of average performance comparisons of different input video frames in the baseline model Resnet101-LSTM, and the main adopted evaluation indexes comprise: accuracy value, AUC value, F1-Score value, recall value (Recall), precision (Precision). In this embodiment, the different input frames and the corresponding model evaluation indexes are visualized, and the visualization result is shown in fig. 5
TABLE 1
As can be seen from table 1 and fig. 5, as the number of input video frames increases, the accuracy of the model tends to increase and then decrease, and at 35 frames, the accuracy is highest and can reach 81.92%. While other corresponding indicators are not quite different. Therefore, the selection of 35 frames of the input video frame is taken as a standard for the subsequent experiments related to the method of the invention.
Then, the present embodiment verifies the improvement of the pain recognition performance by the proposed dynamic fusion module and the comparison of the model complexity and the parameter amount, as shown in tables 2 and 3.
TABLE 2
TABLE 3 Table 3
Wherein, table 2 is the average performance comparison of AFNB, DFNB fusion modules for pain identification; table 3 shows the comparison of the complexity and parameter amounts of different models.
Finally, the facial image dynamic video-based pain identification in the facial image dynamic video-based pain assessment method based on the dynamic fusion module is compared with the most advanced model result in the UNBC-McMaster shoulder pain dataset. It should be noted that, when comparing with other models, in order to ensure consistency, a leave-one method is adopted to divide data, namely, 24 tested data are used for model training in 25 tested data, the remaining tested data are subjected to model test, the cycle is 25 times and each time the data cannot be repeated, and finally, the 25 test results are averaged to obtain corresponding results, and the results are shown in table 4.
TABLE 4 Table 4
In summary, the method of the invention performs corresponding tests on the public data set UNBC-McMaster Shoulder Pain Expression Archive, and the proposed model performance is superior to that of the prior other methods, thereby providing an auxiliary means for clinical pain assessment and having potential application prospects for assessing the pain scene of a special clinical patient.
Those of ordinary skill in the art will recognize that the embodiments described herein are for the purpose of aiding the reader in understanding the principles of the present invention and should be understood that the scope of the invention is not limited to such specific statements and embodiments. Various modifications and variations of the present invention will be apparent to those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the scope of the claims of the present invention.
Claims (4)
1. A facial dynamic video image pain assessment method based on a dynamic fusion module comprises the following specific steps:
s1, carrying out data balance processing on a UNBC-McMaster Shoulder Pain Expression Archive data set;
s2, in a pre-training model Resnet101 network, a dynamic fusion module is obtained based on Non-local framework improvement, and an image space feature extraction model is established;
s3, extracting spatial feature information of each frame of image based on the spatial feature extraction model in the step S2, inputting the extracted feature information into a long-short-time memory network to extract corresponding time sequence information of the facial dynamic video image, further extracting the time sequence information between adjacent frames, realizing recognition of facial pain, and finally realizing automatic evaluation of the pain.
2. The method for evaluating facial dynamic video image pain based on the dynamic fusion module according to claim 1, wherein the step S1 is specifically as follows:
the UNBC-McMaster Shoulder Pain Expression Archive data set comprises 25 tested, 200 videos and 48398 frames of images, and each frame of image is marked with pain degree by using a PSPI method;
the data balancing processing mode specifically comprises the following steps:
(1) Deleting the complete sequence containing the pain-free frames;
(2) If the beginning or ending frame of a complete sequence contains a pain-free frame, then the portion is deleted.
3. The method for evaluating facial dynamic video image pain according to claim 1, wherein in the step S2, the dynamic fusion module is configured to fuse the features extracted by the fourth block and the fifth block of the mesh 101 of the pre-training model, specifically as follows:
features extracted by a fourth block and a fifth block of the pre-training model Resnet101 network are respectively marked as X h 、X l I.e. a high-level feature mapAnd a low-level feature map +.>
Wherein N is h 、N l Respectively represent X h 、X l Spatial position number, C h 、C l Respectively represent X h 、X l Is used for the number of channels of a computer,representing a feature set;
fusing all extracted features of a fourth block and a fifth block in the Resnet101 network, inputting the fused features into a multi-scale channel attention module to calculate a weight value alpha for weighting X h And 1-alpha is used to weight X l Is a feature of the input of (a); the obtained characteristics after weighting by the weight value alpha and 1-alpha are X 'respectively' h 、X′ l The method comprises the steps of carrying out a first treatment on the surface of the And after the dynamic fusion module, extracting the spatial characteristic information of each input image frame.
4. The method for evaluating facial dynamic video image pain based on the dynamic fusion module according to claim 1, wherein in the step S2, the pre-training model divides the data set by using a five-fold cross-validation method, so as to train and test the proposed network model, specifically as follows:
1) Training and testing the ratio of the data sets to be 4:1, training by using an Adam algorithm, updating parameters in a network model and storing the model;
2) And (3) repeating the step (1) for iterative training, and when the training is finished and the evaluation index reaches the highest, the model is called as an optimal model, and the optimal model is stored.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310526165.1A CN116525145A (en) | 2023-05-10 | 2023-05-10 | Facial dynamic video image pain assessment method based on dynamic fusion module |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310526165.1A CN116525145A (en) | 2023-05-10 | 2023-05-10 | Facial dynamic video image pain assessment method based on dynamic fusion module |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116525145A true CN116525145A (en) | 2023-08-01 |
Family
ID=87397431
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310526165.1A Pending CN116525145A (en) | 2023-05-10 | 2023-05-10 | Facial dynamic video image pain assessment method based on dynamic fusion module |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116525145A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109272048A (en) * | 2018-09-30 | 2019-01-25 | 北京工业大学 | A kind of mode identification method based on depth convolutional neural networks |
| US20210030354A1 (en) * | 2018-04-19 | 2021-02-04 | University Of South Florida | Neonatal pain identificaiton from neonatal facial expressions |
| CN113935435A (en) * | 2021-11-17 | 2022-01-14 | 南京邮电大学 | Multi-modal emotion recognition method based on space-time feature fusion |
| CN114943924A (en) * | 2022-06-21 | 2022-08-26 | 深圳大学 | Pain assessment method, system, device and medium based on facial expression video |
-
2023
- 2023-05-10 CN CN202310526165.1A patent/CN116525145A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210030354A1 (en) * | 2018-04-19 | 2021-02-04 | University Of South Florida | Neonatal pain identificaiton from neonatal facial expressions |
| CN109272048A (en) * | 2018-09-30 | 2019-01-25 | 北京工业大学 | A kind of mode identification method based on depth convolutional neural networks |
| CN113935435A (en) * | 2021-11-17 | 2022-01-14 | 南京邮电大学 | Multi-modal emotion recognition method based on space-time feature fusion |
| CN114943924A (en) * | 2022-06-21 | 2022-08-26 | 深圳大学 | Pain assessment method, system, device and medium based on facial expression video |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Multi-label classification of fundus images with efficientnet | |
| CN110399929B (en) | Fundus image classification method, fundus image classification apparatus, and computer-readable storage medium | |
| Zago et al. | Retinal image quality assessment using deep learning | |
| CN109308692B (en) | OCT image quality evaluation method based on improved Resnet and SVR mixed model | |
| CN108464840B (en) | Automatic detection method and system for breast lumps | |
| CN109345538A (en) | A kind of Segmentation Method of Retinal Blood Vessels based on convolutional neural networks | |
| CN112801168B (en) | Lesion area prediction analysis method, system and terminal equipment of tumor image | |
| Zhuang et al. | Breast ultrasound lesion classification based on image decomposition and transfer learning | |
| CN114693961B (en) | Fundus photo classification method, fundus image processing method and fundus image processing system | |
| CN110859624A (en) | Brain age deep learning prediction system based on structural magnetic resonance image | |
| CN113012163A (en) | Retina blood vessel segmentation method, equipment and storage medium based on multi-scale attention network | |
| CN115035127B (en) | Retina blood vessel segmentation method based on generation type countermeasure network | |
| CN115222688B (en) | Medical image classification method based on graph network time sequence | |
| Dai et al. | A novel meibomian gland morphology analytic system based on a convolutional neural network | |
| CN112768065B (en) | Facial paralysis grading diagnosis method and device based on artificial intelligence | |
| CN114926396B (en) | Mental disorder magnetic resonance image preliminary screening model construction method | |
| CN110610480B (en) | MCASPP neural network eyeground image optic cup optic disc segmentation model based on Attention mechanism | |
| CN112036298A (en) | Cell detection method based on double-segment block convolutional neural network | |
| CN115330748A (en) | Brain image processing method, computer device, and storage medium | |
| CN114565601A (en) | Improved liver CT image segmentation algorithm based on DeepLabV3+ | |
| CN110020597B (en) | Eye video processing method and system for auxiliary diagnosis of dizziness/vertigo | |
| CN113160151A (en) | Panoramic film dental caries depth identification method based on deep learning and attention mechanism | |
| CN116525145A (en) | Facial dynamic video image pain assessment method based on dynamic fusion module | |
| CN117338234A (en) | Diopter and vision joint detection method | |
| CN116092667A (en) | Disease detection method, system, device and storage medium based on multi-mode images |
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 |