NCNN+Int8+YOLOv4 quantitative modeling and real-time inference
Int8 quantization and inference for yolov4-tiny using ncnn is performed as follows.
- inference
| Equipment | Computing backend | System | Framework | input_size | Run time |
|---|---|---|---|---|---|
| Intel | Core i5-4210 | window10（x64） | ncnn@fp16 | 320 | 36ms |
| Intel | Core i5-4210 | window10（x64） | ncnn@int8 | 320 | 57ms |
| Raspberrypi 3B | 4xCortex-A53 | Linux(arm64) | ncnn@fp16 | 320 | 313ms |
| Raspberrypi 3B | 4xCortex-A53 | Linux(arm64) | ncnn@int8 | 320 | 217ms |
-
Installing NCNN
git clone https://github.com/Tencent/ncnn.git cd ncnn mkdir build cd build cmake .. make make install
中文教程：https://zhuanlan.zhihu.com/p/368653551
-
Run Demo
git clone https://github.com/pengtougu/ncnn-yolov4-int8.git Open VS-2019 and create project copy yolov4.cpp run
中文教程：https://zhuanlan.zhihu.com/p/372278785
- Note
Please watch out the following code changes
