PaddleX/paddlex/configs/modules/open_vocabulary_detection/GroundingDINO-T.yaml

Global:
  model: GroundingDINO-T
  mode: predict # only support predict
  device: gpu:0
  output: "output"

Predict:
  batch_size: 1
  model_dir: "/path/to/GroundingDINO-T"
  input: "https://paddle-model-ecology.bj.bcebos.com/paddlex/imgs/demo_image/open_vocabulary_detection.jpg"
  prompt: "bus . walking man . rearview mirror ."
  kernel_option:
    run_mode: paddle