interactive_demo

GAN Compression Interactive Demo on Jetson Nano

Overview

We show how to deploy our compressed Pix2pix model on NVIDIA Jetson Nano. Our model is compiled with TVM Auto Scheduler [1,2] for acceleration. The final model achieves about 8 FPS on Jetson Nano GPU.

[1] Tianqi Chen et al., TVM: An automated end-to-end optimizing compiler for deep learning, in OSDI 2018

[2] Lianmin Zheng et al., Ansor: Generating High-Performance Tensor Programs for Deep Learning., in OSDI 2020

Getting Started

1. Get an NVIDIA Jeston Nano board (it is only $99!).

2. Get a micro SD card and burn the Nano system image into it following here. Insert the card and boot the Nano. Note: you may want to get a power adaptor for a stable power supply.

3. Follow here to install PyTorch and torchvision.

4. Install TVM 0.8.

   # upgrade cmake
   cd ~
   sudo apt install openssl libssl-dev
   sudo apt remove cmake
   wget https://github.com/Kitware/CMake/releases/download/v3.21.0-rc2/cmake-3.21.0-rc2-linux-aarch64.sh
   sudo bash cmake-3.21.0-rc2-linux-aarch64.sh --prefix=/usr --exclude-subdir --skip-license
   
   cd ~
   sudo apt install llvm # install llvm which is required by tvm
   git clone --recursive https://github.com/apache/tvm tvm
   cd tvm
   mkdir build
   cp cmake/config.cmake build/
   cd build
   vim config.cmake
   # edit config.cmake to change
   # USE_CUDA OFF -> USE_CUDA ON
   # USE_LLVM OFF -> USE_LLVM ON
   # USE_THRUST OFF -> USE_THRUST ON
   # USE_GRAPH_EXECUTOR_CUDA_GRAPH OFF -> USE_GRAPH_EXECUTOR_CUDA_GRAPH ON
   
   cmake ..
   make -j4
   cd ..
   cd python; python3 setup.py install --user; cd ..

5. Install PyQt5 (Jetson Nano may have pre-installed it).

6. You now can try our demo with

   python3 paint.py

Acknowledgement

This demo code is developed based on Piecasso.