This repo contains artifacts of the TensorCV project.
OS: Ubuntu 20.04 LTS
CPU: Intel Core i9 12900KF
GPU: Nvidia RTX 3090
GCC: 9.4.0
CUDA: 11.7
We recommend you to use the docker image nvidia/cuda:11.7.0-devel-ubuntu20.04 (https://hub.docker.com/r/nvidia/cuda)
docker run -itd --name <name> --gpus all nvidia/cuda:11.7.0-devel-ubuntu20.04
docker exec -it <name> bash
Setup environment
apt update
apt-get install -y build-essential git cmake make
Install CUDA (https://developer.nvidia.com/cuda-downloads)
Install opencv and opencv_contrib with cuda support (https://opencv.org/)
git clone https://github.com/opencv/opencv.git
git clone https://github.com/opencv/opencv_contrib.git
cd opencv && mkdir build && cd build
cmake -DOPENCV_EXTRA_MODULES_PATH=../../opencv_contrib/modules -DWITH_CUDA=ON ..
cmake --build . -- -j && make install -j
Build TensorCV
git clone https://github.com/escalab/TensorCV.git
cd TensorCV && mkdir build && cd build
cmake ..
make
./cvTest CV 4032 # Run TensorCV part only with 4032x3024 input images
./cvTest <MODE> <SIZE>
MODE: CPU/GPU/CV/ALL
SIZE: 480 / 1600 / 2048 / 2592 / 3264 / 4032
This repository only provides 20 images as samples
| SIZE | Input image size (directory) |
|---|---|
480 |
480 x 320 (img/480) |
1600 |
1600 x 1200 (img/1600) |
2048 |
2048 x 1536 (img/2048) |
2592 |
2592 x 1936 (img/2592) |
3264 |
3264 x 2448 (img/3264) |
4032 |
4032 x 3024 (img/4032) |