Added Dockerfile for CUDA compatibility, including test file

[teenaxta]

Set cuda architecture to ensure that docker file compiles with CUDA on the latest pytorch docker image

[anas-zafar]

@teenaxta could you please share instructions how to run this model in docker? Thanks

[snenyl]

For people that run RTX 3060.

Dockerfile.txt

Build with the following command

docker build -t groundingdino:dev .

Run and mount with selected <host-image-folder>
docker run --gpus all -it --rm -v <host-image-folder>:/GroundingDINO/images groundingdino:dev /bin/bash

While inside the docker container run the following command

CUDA_VISIBLE_DEVICES=0 python3 demo/inference_on_a_image.py \
    --config_file groundingdino/config/GroundingDINO_SwinT_OGC.py \
    --checkpoint_path weights/groundingdino_swinb_cogcoor.pth \
    --image_path /GroundingDINO/images/store.webp \
    --text_prompt "plastic" \
    --output_dir /GroundingDINO/images/output \
    --box_threshold 0.35 \
    --text_threshold 0.25