Course project for SJTU CS385: Machine Learning, taught by Prof. Quanshi Zhang
- numpy
- tensorflow
Unzip FDDB-folds.tgz
and originalPics.tar.gz
into data/
Run python preprocess.py
To do face recognization, run python recognization.py
with following arguments
-model: model name, lr/svm/fisher/cnn
-lr: learning rate
-total_epoches: total training epoches
-batch_size: batch size for cnn model
-kernel: kernel type of svm
-langevin: if set, use langevin dynamics
-gpu: gpu device
To do face detection, run python detection.py
with following arguments
-model: model name, lr/svm/cnn
-winSize: sliding window size
stride: sliding window stride length
thres_score: threshold of score
thres_iou: threshold of IoU
scales: scale factors, use , to seperate
hw_ratios: height-width ratios, use , to seperate