Closed fblogy closed 2 years ago
You can output boxes from all classes and only select those from the known class.
Are you replace the all unknown class embedding by label embedding of class c in the matching part ,and the number of times you need to inference is the number of classes in the image?
Yes, your understanding is right.
Thanks for your excellent work. May I ask about the specific implementation of the fine-tuning phase?
Thanks for your excellent work. Could you give more details how Known Labels Detection implemented?How do you let the decoder output all boxes of specific class c only using the label embedding of class c?