cascade-detr
cascade-detr copied to clipboard
SysCV
[ICCV'23] Cascade-DETR: Delving into High-Quality Universal Object Detection
Great job! I have some questions about the UVO dataset. Did you directly use the official 'video2frames.py' script from the website to convert videos into images? Because the paper mentions...
Hi, I appreciate your work on Cascade DETR. However, I am a bit confused by the description of the **predicted bounding boxes** in your statement: "Si is the set of...
In the caption of Figure 5, you said "We then compute the IoU with ground truth for each prediction and show a cumulative average of IoU." How is the correspondence...
你好，我想问一下，论文中为什么说detr的query scoring 是存粹基于预测的分类置信度？采用分类置信度作为对所有查询建议进行排序的代理？detr不是根据预测的100个box的类别和边界框的损失，然后计算匈牙利损失矩阵，然后进行匹配，得到相应的匹配结果。根据匹配的结果计算最终所有损失的加权和，反向梯度回传，更新参数。query scoring 具体是指什么？
Want to use it as a graduation thesis baseline
Great job! I have some questions regarding the EgoHands dataset. 1. In Table 1 of the paper, the EgoHands dataset is mentioned to contain 11k images, but the dataset I...
