CUDANet/examples/inception_v3

Inception v3

Inception v3 Inference on CUDANet

Usage

1. Export pytorch Inception v3 weights pretrained on ImageNet (requires pytorch and torchvision):

python inception_v3.py

2. Follow the instructions from the repository root to build the CUDANet library.

3. Build Inception v3 (requires OpenCV for image loading and preprocessing):

mkdir build
cd build
cmake -S ..
make

4. (Optional) Run tests

Generate test input/output and resources by running inception_blocks.py in the test folder

Build and run tests (requires Google Test)

cd build
make test_inception_v3
./tests/test_inception_v3

5. Run Inception v3 inference:

inception_v3 ../inception_v3_weights.bin ../image.jpg

Note on Preprocessing

The image preprocessing in this implementation uses OpenCV, which may produce slightly different results compared to PyTorch's Pillow-based preprocessing due to differences in interpolation methods during resizing.