智慧教室
课堂专注度及考试作弊系统、课堂动态点名,情绪识别、表情识别和人脸识别结合
课堂专注度分析
课堂专注度+表情识别
作弊检测
关键点计算方法
转头(probe)+低头(peep)+传递物品(passing)
侧面的传递物品识别
逻辑回归关键点
下载权重
Halpe dataset (136 keypoints)
1.Model | Backbone | Detector | Input Size | AP | Speed | Download | Config | Training Log |
---|---|---|---|---|---|---|---|---|
Fast Pose | ResNet50 | YOLOv3 | 256x192 | 69.0 | 3.54 iter/s | Google Baidu | cfg | log |
- 放到detection_system/checkpoints
2. Human-ReID based tracking (Recommended)
Currently the best performance tracking model. Paper coming soon.
Getting started
Download human reid model and place it into AlphaPose/trackers/weights/
.
Then simply run alphapose with additional flag --pose_track
You can try different person reid model by modifing cfg.arch
and cfg.loadmodel
in ./trackers/tracker_cfg.py
.
If you want to train your own reid model, please refer to this project
3. Yolo Detector
Download the object detection model manually: yolov3-spp.weights(Google Drive | Baidu pan). Place it into detector/yolo/data
.
4. face boxes 预训练权重
- 放到face_recog/weights文件夹下
5. 其他
百度云 提取码:rwtl
人脸识别:dlib_face_recognition_resnet_model_v1.dat
- detection_system/face_recog/weights
人脸对齐:shape_predictor_68_face_landmarks.dat
- detection_system/face_recog/weights
作弊动作分类器:cheating_detector_rfc_kp.pkl
- detection_system/weights
使用
运行setup.py安装必要内容
python setup.py build develop
运行demo_inference.py
将detection_system设置为source root
使用摄像头运行程序
python demo_inference.py --vis --webcam 0