IDEA-Research / T-Rex

[ECCV2024] API code for T-Rex2: Towards Generic Object Detection via Text-Visual Prompt Synergy
https://deepdataspace.com/blog/T-Rex
Other
2.28k stars 147 forks source link

only one class visual prompt for once #41

Closed GallonDeng closed 7 months ago

GallonDeng commented 7 months ago

for generic visual prompt mode, it only supports one class prompt, right? It is possible to detect different classes simultaneously with multiple prompts for different classes?

Mountchicken commented 7 months ago

Hi @AllenDun This feature is supported for sure. The essence of multi-category inference is to have a visual prompt embedding for each category. You can use customize_embedding multiple times to customize the embedding for each category. and then use embedding_inference for multi-category inference.

GallonDeng commented 7 months ago

@Mountchicken thanks for reply. So it is supported when use API, not the online demo

Mountchicken commented 7 months ago

Yes