MTCNN-Pytorch:人脸检测模型

# Instructions MTCNN is a face detection model The Chinese introduction can be found [here](https://blog.csdn.net/qq_32478489/article/details/106193921) [paper](https://kpzhang93.github.io/MTCNN_face_detection_alignment/paper/spl.pdf) Referring to [kuaikuaikim/DFace](https://github.com/kuaikuaikim/DFace) and [Sierkinhane/mtcnn-pytorch](https://github.com/Sierkinhane/mtcnn-pytorch), I fixed some bugs that would appear in the training, added the table of learning rate, and optimized the training parameters.Testing on the WiderFace validation set, I found it performance better than the model weight they had originally provided. #### Test image  # WiderFace Val Performance in MTCNN MTCNN-original is the test result of the original weight parameter MTCNN-trained is the test result of my training | Style | easy | medium | hard | |:-|:-:|:-:|:-:| | MTCNN-original | 65.3% | 65.1% | 40.3% | | MTCNN-trained | **71.4%** | **70.4%** | **43.2%** |    # Installation 1.pytorch 2.opencv # test ### a single picture Modify the image path in the program ```Shell python detect.py ``` # Train the model download [widerface](http://shuoyang1213.me/WIDERFACE/) Organise the dataset directory as follows: ```Shell ./data_set/face_detection/ WIDER_train/ images/ WIDER_val/ images/ ``` I have made the label file ```Shell ./anno_store/anno_train.txt ``` If you want to make your own labels The reference program ./anno_store/tool/change.py ### train pnet **prepare pnet data** ```Shell python mtcnn/data_preprocessing/gen_Pnet_train_data.py python mtcnn/data_preprocessing/assemble_pnet_imglist.py ``` **train pnet** ```Shell python mtcnn/train_net/train_p_net.py ``` **prepare rnet data** ```Shell python mtcnn/data_preprocessing/gen_Rnet_train_data.py python mtcnn/data_preprocessing/assemble_rnet_imglist.py ``` **train rnet** ```Shell python mtcnn/train_net/train_r_net.py ``` **prepare onet data** ```Shell python mtcnn/data_preprocessing/gen_Onet_train_data.py python mtcnn/data_preprocessing/assemble_onet_imglist.py ``` **prepare onet** ```Shell python mtcnn/train_net/train_o_net.py ``` # Evaluation widerface val ```Shell python wildface_test.py ``` # References [kuaikuaikim/DFace](https://github.com/kuaikuaikim/DFace) [Sierkinhane/mtcnn-pytorch](https://github.com/Sierkinhane/mtcnn-pytorch)