FaceBoxes is a cpu real-time face detector with high accuracy
ncnn is a high-performance neural network inference computing framework optimized for mobile platforms.
repair some bug in onnx2ncnn
1.onnx2ncnn in ncnn don't support the operation which changes Split to Slice. Modifying the onnx2ncnn on the basis of ncnn in order to support Split to Slice operator.
2.onnx2ncnn in ncnn would parse to the wrong parameter in reshape operator if reshape operator is 4 dims(eg:torch.reshape(b, c, h, w) to ncnn reshape).
3.winograd3x3 conv style has bug, I don't use winograd3x3.
-
set opencv path in ./faceboxes/CMakeLists.txt file
-
cd ./build/faceboxes/ && ./facebox, test.jpg saved in ./model/test.jpg
We provide the converted model in model folder!
1.ncnn
2.facebox