Product Details 商品属性
- Language: Python
- Database: 无
- License: License-Free
- Source Code: Fully Open Source
- Specification: Full Website Source
- Demo URL: https://www.bilibili.com/video/BV1M3411r7gE/
- Delivery: Automatic (instant download after payment)
Demo Video:
You can go to station B to watch. https://www.bilibili.com/video/BV1M3411r7gE/。
Behavioral safety monitoring
In order to solve the problems of insufficient accuracy and low efficiency in the current behavior monitoring, the posture recognition technology of OpenPose is combined to build a classification algorithm through the coordination relationship between different limbs, and through the comparison of different classification algorithms, the optimal model is selected to build a multi-objective classification method, and finally the real-time display of posture display, target detection and classification of multiple targets can be realized. In this model, the human pose recognition is carried out by calling the lightweight OpenPose model, and the main method is to obtain the position of each bone key point of the human body through OpenPose, and then match the two bones through the Euclidean distance to specifically detect each person, and the missing bone key points in the common detection can be filled by the bone information of the previous frame.
Fitness scoring system
Using the pose estimation technology, after extracting the key points of the human body, the different human limbs are combined, the bone vector angle data is evaluated in real time, and the score of each limb and the total score are given according to the comparison of the human body vector angle data with the established standard movements. Here we take dumbbell lifting as an example to build our smart fitness scoring system.
Posture correction system
Through attitude estimation and then cosine vector angle calculation, attitude contrast correction is realized
Introduction to the code
Technology stack: Python language, PyQt5 to establish a GUI interface, OpenPose to implement human pose estimation, Keras to implement model training.
Environmental requirements can be found in the blog: https://blog.csdn.net/qq_42279468/article/details/124987801?spm=1001.2014.3001.5502
Algorithms: OpenPose, a human pose estimation algorithm, a vector angle comparison algorithm for bone angle extraction, and a neural network multi-classification algorithm
Technical documentation display
The training dataset is displayed
Demonstration of the training process
(more…) ）
