FaceDetection_CNN

Implement Yahoo Paper: Multi-view Face Detection Using Deep Convolutional Neural Networks

1. Image Preprocess aflw dataset[1]. Use iou>=0.5 as positive, iou<=0.3 as negative. You should download the aflw dataset by yourself.

2. Fine-tune Alex-Net using AFLW dataset. The model is in Baidu Yun: https://pan.baidu.com/s/1pJJ2WKN, or Google Drive: https://drive.google.com/file/d/0B8_dH3SiT7reMjJVRjJDXzJkRDQ/view?usp=sharing

3. Convert fully connected layers into convolutional layers by reshaping layer parameters, see [2], you can use the convert_full_conv() function in test.py for converting.

4. Get heat map for each scale of image.

5. Process heat map by using non-maximal suppression to accurately localize the faces.

========== Reference: [1]https://lrs.icg.tugraz.at/research/aflw/

[2] http://nbviewer.ipython.org/github/BVLC/caffe/blob/master/examples/net_surgery.ipynb

##TODO The image preporcessing creation has the following defienceies

1. Doesn't create as many as images as it was reported in the paper(arround 20.2 Million images)
2. Doesn't really avoid overlaping faces to assume non overlap. Needs to be fixed.
3. Isn't Threaded till NOW.