tao-train-mask-grounding-dino

Automate TAO Mask-Grounding-Dino model training and inference by using a workflow that handles data preparation, validation, and execution. This requires Docker and the NVIDIA Container Toolkit to run the necessary TAO tools for open-set instance segmentation tasks.

Export a trained instance segmentation model to ONNX format using the model export functionality provided in the TAO workflow. This allows you to take the trained Mask-Grounding-Dino model and prepare it for deployment and inference on new data.

Yes, you need Docker and the NVIDIA Container Toolkit to run TAO training automation. These dependencies are required to execute the specific TAO tools that handle the training, evaluation, and inference processes for Mask-Grounding-Dino models.

Evaluate open-set instance segmentation models by running them against a set of predefined tasks during the validation phase. This ensures the trained Mask-Grounding-Dino model's performance meets the required criteria before you proceed to export and deployment.

Yes, you can use TAO Mask-Grounding-Dino for grounded instance segmentation on new data. The workflow provides a dedicated inference mechanism that applies your trained and exported models to process and segment novel visual inputs.

Your TAO model export or training workflow requires specific environment tools because the automation relies on Docker and the NVIDIA Container Toolkit to isolate dependencies and execute the specialized TAO commands needed for open-set instance segmentation.