ncnn is a high-performance neural network inference computing framework optimized for mobile platforms.
how to use ncnn with alexnet, and how to convert caffe model to ncnn model
Follow these steps after you build the library using above compile instructions.
You only need .param files in ncnn format to test inference performance. Copy all the param files from NCNNSFD/benchmark/ to NCNNSFD/build/benchmark/
$ cd build/benchmark/
$ ./benchcnn 8 4 0
---Output---
Copy .jpg images from NCNNSFD/ncnnsfd/ to NCNNSFD/build/ncnnsfd/. Copy .param and .bin files from NCNNSFD/ncnnsfd/ to NCNNSFD/build/ncnnsfd/.
- Test in single image.
$ cd NCNN/build/ncnnsfd/
$ ./mobilenetsfd_image 79378097.jpg NVISO-A7.param NVISO-A7.bin
---Output---
- Test in webcam stream.
$ cd NCNN/build/ncnnsfd/
$ ./mobilenetsfd_webcam 0 NVISO-A7.param NVISO-A7.bin
---Output--- ...
BSD 3 Clause