Request for Access Review and Reconsideration – SAM 3
Hello SAM 3 Team,
I recently requested access to the official SAM 3 model weights through Hugging Face, but unfortunately my request was rejected.
I am conducting academic research on Test-Time Adaptive Object Detection (TTAOD), particularly exploring how independent vision foundation models can help evaluate object detection quality without ground-truth annotations.
I am interested in using SAM 3 as an independent visual verifier. Specifically, I would like to investigate whether its text-prompted segmentation capabilities can provide independent instance-level evidence to evaluate and improve the ranking of object detection candidates generated by Grounding DINO.
My intended research involves:
- Using SAM 3 with full images and category text prompts.
- Evaluating the consistency between independently generated segmentation masks and object detection bounding boxes.
- Investigating whether this information can improve object detection quality assessment.
I fully respect Meta's access requirements and the model's licensing terms.
Could you please clarify whether my access request can be reconsidered? If additional information is required, or if there is a procedure to reset my rejected request and allow me to apply again, I would be grateful for your guidance.
Thank you for your time and consideration.
Best regards,
An academic researcher