tao-run-deft-object-detection
tao-run-deft-object-detection is an agent skill published by nvidia. It lives in the NVIDIA/skills repository, which has 3,735 GitHub stars and is licensed under Apache-2.0; we last read it on 2026-10-18.
Runs the DEFT smart-data-augmentation loop for NVIDIA TAO Grounding DINO object detection, including zero-shot inference, KPI and per-class gap analysis, SigLIP embedding, unique-neighbor mining, and retraining, repeated for a fixed number of iterations. Prepares the source pool used for mining, involving Co-DETR pseudo-labeling and dataset conversions. Requires credentials for TAO and DEFT.
Install
skills is a generic installer for any repository of Claude agent skills -- it reads NVIDIA/skills’s own structure rather than assuming one, the same way this page reads the skill’s own SKILL.md rather than assuming a format.
npx skills add NVIDIA/skills
What it connects to
Unverified means a mention in the skill’s own description, not a citation. A wrong claim here costs nothing to correct and nothing to trust yet.
Maintenance
Run the full DEFT smart-data-augmentation loop for NVIDIA TAO Grounding DINO object detection: zero-shot baseline inference, KPI analysis, per-class gap analysis, SigLIP embedding of weak images, unique-neighbor mining against a source pool, ODVG dataset staging, and retraining — repeated for a fixed number of iterations. Also prepares the source pool the loop mines from, as a separate run: Co-DETR pseudo-labeling, folding to the target classes, KITTI→COCO→ODVG conversion, and embedding. Use for prompts like "run the DEFT OD loop", "run smart data augmentation for grounding dino", "mine and retrain my detection model", "improve OD mAP with gap analysis and mining", "prep the source pool", or "pseudo-label my unlabeled images for mining"; do not use for standalone TAO training, one-off inference, or gap analysis alone.
Common questions about tao-run-deft-object-detection
- What does tao-run-deft-object-detection do?
- Runs the DEFT smart-data-augmentation loop for NVIDIA TAO Grounding DINO object detection, including zero-shot inference, KPI and per-class gap analysis, SigLIP embedding, unique-neighbor mining, and retraining, repeated for a fixed number of iterations. Prepares the source pool used for mining, involving Co-DETR pseudo-labeling and dataset conversions. Requires credentials for TAO and DEFT.
- How do I install tao-run-deft-object-detection?
- Run npx skills add NVIDIA/skills. The skills installer reads the repository’s own structure, so it works for any repository of agent skills.
- Who publishes tao-run-deft-object-detection?
- nvidia.
- Is tao-run-deft-object-detection maintained, and what licence does it use?
- Its repository has 3,735 GitHub stars and its state in our directory is listed. Its last commit we recorded was on 2026-10-26. The repository declares the Apache-2.0 licence.