增加了多类别同时识别
Added simultaneous recognition of multiple categories
本程序基于grouding DINO 和 segment anything model(SAM)。
This program is based on the grounding DINO and segment anything model (SAM).
20230512_192253.mp4
支持多种图片格式,如'.jpg','.png','.bmp','.jpeg'
Supports multiple image formats, such as' *. jpg ',' *. png ',' *. bmp ',' *. jpeg '
Clone the GroundingDINO repository from GitHub.
git clone https://github.com/IDEA-Research/GroundingDINO.git
Change the current directory to the GroundingDINO folder.
cd GroundingDINO/
Install the required dependencies in the current directory.
pip3 install -q -e .
clone the Segment Anything repository locally and install with
git clone git@github.com:facebookresearch/segment-anything.git
cd segment-anything; pip install -e .
install labelme
pip install labelme
注意
你需要一些依赖但是目前我还没有整理出来
You need some dependencies, but I haven't sorted them out yet