- Clone this repo.
- Put your Imagenet validation set in
val_files/
.
./val_files/
├── ILSVRC2012_val_00000001.JPEG
├── ILSVRC2012_val_00000002.JPEG
├── ILSVRC2012_val_00000003.JPEG
├── ILSVRC2012_val_00000004.JPEG
├── ILSVRC2012_val_00000005.JPEG
├── ILSVRC2012_val_00000006.JPEG
...
- Pull nvcr.io/nvidia/tensorrt:19.09-py3 from NGC.
- Start nvcr.io/nvidia/tensorrt:19.09-py3 container.
- Set environment in host (Modify the script according to your situation):
./set_env.sh
- In the container, run cmd as below:
# Install opencv
apt-get update
apt-get install libopencv-dev
# Assuming you mounted this repo to /dawnbench
cd /dawnbench/
# Build
./make.sh
# Run
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PWD
./inference engine/engine.plan val_files/ labels.txt