Vision Language Mapping

This repo uses clipseg for semantic segmentation and create a costmap using lidar pointclouds

Installation

Use the package manager pip to install dependencies.

pip install -r requirements.txt

Usage

To run segmentation from a folder containing images.s

python run_folder.py --image_folder /path/to/images --prompts "prompt_1" "prompt_2" ... "prompt_n" --output_folder /path/to/output

To run segmentation from a rosbag.

python run_bag.py --bag_file_path /path/to/rosbag --prompts "prompt_1" "prompt_2" ... "prompt_n" --topic <image_topic>