Runsen Xu

I have the same question.

Hi, there is a colorToDepthExtrinsics in each {scene_id}.txt. Have you used that? However, according to https://github.com/ScanNet/ScanNet/issues/92, they are already aligned and no extrinsic matrix should be applied. I am a...

Dear Liu, If I want to reproduce LLaVA, which one am I recommended to use, Vicuna-v1.1 or Vicuna-V0? The repo supports both VicunaV0 and VicunaV1.1, right? Also, which model are...

I see. Thank you very much!

I have the same question here. Actually, we can use the bounding box or center point to prompt sam to get the mask at the first frame. But usually, the...

Hi, can I draw a conclusion that in the Ultralytics codebase, when models accept a numpy array as an image, they assume the array has RGB format? Then, is this...

I am quite confused now. After checking the loader codes, it seems that: 1. If passing the image path, then models use `LoadImagesAndVideos` for loading images. However, this dataset load...

> This actually doesn't clear up things, because as seen in the example codes, the following produce identical results: > > ``` > image_fn = 'path/to/image.jpg' > im_bgr = cv2.imread(image_fn)...

There is indeed another pre-processing module before the model forward that changes BGR to RGB again. See below: https://github.com/ultralytics/ultralytics/blob/fcfc44ea9cf529685e0afaba6e62f7d2d40c84d8/ultralytics/engine/predictor.py#L123-L127 So to summarize, for the model forward process, it assumes RGB...