Face Detection MTCNN Pytorch로 학습하기

1. 아나콘다 환경세팅

2. 데이터셋 다운로드 후 dataset폴더생성

 

https://deepai.org/dataset/wider-face

 

3. run > python ./anno_store/tool/format/transform.py change .mat(wider_face_train.mat) into .txt(anno_train.txt)

 

아래 경로 확인하기

path_to_image

file_to_label

target_file

 

 

4. P-Net 학습시키기

 

prefix = ''
anno_file = "./anno_store/anno_train.txt"
im_dir = "./data_set/face_detection/WIDERFACE/WIDER_train/WIDER_train/images"
pos_save_dir = "./data_set/train/12/positive"
part_save_dir = "./data_set/train/12/part"
neg_save_dir = './data_set/train/12/negative'

 

경로확인

 

python mtcnn/data_preprocess/gen_Pnet_train_data.py

python mtcnn/data_preprocess/assemble_pnet_imglist.py

python mtcnn/train_net/train_p_net.py

 

5. R-Net 학습시키기

 

python mtcnn/data_preprocess/gen_Rnet_train_data.py

 

python mtcnn/data_preprocess/assemble_rnet_imglist.py

python mtcnn/train_net/train_r_net.py

 

6. O-Net 학습시키기

 

python mtcnn/data_preprocess/gen_Onet_train_data.py

python mtcnn/data_preprocess/gen_landmark_48.py

 

 

 

python mtcnn/data_preprocess/assemble_onet_imglist.py

 

 

python mtcnn/train_net/train_o_net.py

 

7. 테스트하기

 

python mtcnn_test.py