Skip to content

Latest commit

 

History

History
44 lines (33 loc) · 1.87 KB

README.md

File metadata and controls

44 lines (33 loc) · 1.87 KB
Raw

Mask-Removal-AI

dataset not pushed

The original kaggle csv dataset

i put https://www.kaggle.com/datasets/sbaghbidi/human-faces-object-detection/data in a folder named faceDataset

it consists of a faceDataset/faces.csv and a images directory with 3000+ images

Face mask yolo dataset

i downloaded another dataset https://universe.roboflow.com/jonathan-chandra/face-mask-object-detection from detecting face mask. it has a yolo format so i need to call its directory datasets. i want to steal a pre-trained yolo model but i only got a dataset.

The kaggle yolo dataset

csv_to_yolo.py

i use csv_to_yolo.py to convert faceDataset/faces.csv into labels of yolo format so we can use those later? The labels are stored in faceDataset/labels

add_face_mask.py

i use add_face_mask.py to add face masks (face_mask.png) into the images of faceDataset their file names ends with _modified

yolo model

face_detection.py

i trained the yolo model with face_detection.py for 8 epochs (i found that i forgot to load the previous trained model for the first few models :P) and it automatically stored the resulting model in runs/detect/train10/weights/best.pt

need to change the model inside the code

detect.py

to see the result, run detect.py

need to change the model and image inside

note: i put: 00000xxx.jpg 00000xxx_modified.jpg to test 00001xxx.jpg 00002xxx.jpg 00001xxx_modified.jpg 00002xxx_modified.jpg to train 00003xxx.jpg 00003xxx_modified.jpg to valid for better training since train11 rip 30 minute a epoch on my labtop

how to use colab to train?

TODO

  • pass family photo to detect.py
  • if masked, do similarity checker with other photos provided by user
  • image registration inside each bounding boxes (similar to how i add face masks in add_face_mask.py?)
  • output final image