georgia-tech-db / visualroad

Visual Road: A Video Data Management Benchmark

Visual Road: A Video Data Management Benchmark

• This repo is an extension of visualroad, please check their project website for more details about the benchmark, links to the papers, sample videos, and pregenerated datasets.

Building the Docker Image

1. Install Docker CE, if not already installed.
2. Install Python 3.6 or later, if not already installed:
3. Clone the Visual Road repository and build the benchmark image:

git clone https://github.com/georgia-tech-db/visualroad
cd visualroad
docker build -t carlasim/eva .

Synthetic Dataset Generation

1. Generate a scale-one dataset with videos of 10 sec each with the following command. Note: -l flag specifies what objects you need ground truth info for.

mkdir dataset
chmod 777 dataset
docker-compose run generator -s 1 -d 10 -l [car,pedestrian] dataset

2. The generator service supports a number of additional options (e.g., --height, --width, --duration). Execute docker-compose run generator -h for a complete list.