search

IDEA-Research / DINO-X-API

DINO-X: The World's Top-Performing Vision Model for Open-World Object Detection and Understanding
https://arxiv.org/abs/2411.14347
Apache License 2.0
366 stars 16 forks source link

hand keypoints detection #6

Open ccchen091 opened 3 days ago

ccchen091 commented 3 days ago

Hello, I got the following error when using your API for hand key point detection, which is supposed to be caused by the model not detecting it. I would like to ask how this should be solved? The image I entered was a jpg format, which must contain the hands.

Traceback (most recent call last): File "/data/d1/chenhao/code/DINO-X-API-main/demo.py", line 76, in detections = sv.Detections( File "", line 9, in init File "/home/chenhao/anaconda3/envs/DINO-API/lib/python3.9/site-packages/supervision/detection/core.py", line 130, in post_init validate_detections_fields( File "/home/chenhao/anaconda3/envs/DINO-API/lib/python3.9/site-packages/supervision/validators/init__.py", line 120, in validate_detections_fields validate_xyxy(xyxy) File "/home/chenhao/anaconda3/envs/DINO-API/lib/python3.9/site-packages/supervision/validators/init__.py", line 11, in validatexyxy raise ValueError( ValueError: xyxy must be a 2D np.ndarray with shape (, 4), but got shape (0,)

rentainhe commented 3 days ago

@ccchen091 Hello, we are still refining our API usage. And I think the issue you met is related to the visualization process, would you like to check if there are predicted boxes or not.

ccchen091 commented 3 days ago

@ccchen091 Hello, we are still refining our API usage. And I think the issue you met is related to the visualization process, would you like to check if there are predicted boxes or not. I went to print out the "predictions" and found that the "predictions" were empty, which should be caused by the API not detecting the hand in the image, but I don't know why the model did not detect the hand in the image.

rentainhe commented 3 days ago

@ccchen091 Hello, we are still refining our API usage. And I think the issue you met is related to the visualization process, would you like to check if there are predicted boxes or not. I went to print out the "predictions" and found that the "predictions" were empty, which should be caused by the API not detecting the hand in the image, but I don't know why the model did not detect the hand in the image.

would you like to provide us with your image and text prompt, which may be more convenient for us to solve this issue.

ccchen091 commented 3 days ago

@ccchen091 Hello, we are still refining our API usage. And I think the issue you met is related to the visualization process, would you like to check if there are predicted boxes or not. I went to print out the "predictions" and found that the "predictions" were empty, which should be caused by the API not detecting the hand in the image, but I don't know why the model did not detect the hand in the image.

would you like to provide us with your image and text prompt, which may be more convenient for us to solve this issue.

I'm very sorry. I found out that my text prompt was set incorrectly, and when I set the text prompt to "hand", it was able to detect correctly. Thank you for this excellent work. Thank you for your patience.