LocalLLaMA

1 readers

1 users here now

Community to discuss about Llama, the family of large language models created by Meta AI.

founded 10 months ago

MODERATORS

communick@poweruser.forum

LLM for object detection/labelling (alien.top)

submitted 10 months ago by takezo07@alien.top to c/localllama@poweruser.forum

6 comments fedilink hide all child comments

Hello,

I'm looking for an alternative to Google Vision AI (LABEL_DETECTION, OBJECT_LOCALIZATION) and Amazon Rekognition (DetectLabels).
Any ideas?

Thanks!

top 6 comments

sorted by: hot top controversial new old

[–] Specialist_Ice_5715@alien.top 1 points 10 months ago (1 children)

You'll have to go multi-modal. The best is now fuyu but that's not commercially usable.

[–] takezo07@alien.top 1 points 10 months ago (1 children)

I found Blip: https://replicate.com/salesforce/blip?input=form&output=preview
But that's not exactly what i'm looking for. It does image captioning very well.
Like in the their example: "a woman sitting on the beach with a dog".
But i need a list of objects and "things" like : dog, woman, beach, wave, shirt...etc.

[–] Specialist_Ice_5715@alien.top 1 points 10 months ago

interesting.. is blip commercially usable? I read that it is, but is this correct for the weights in their entirety?

[–] hurrytewer@alien.top 1 points 10 months ago

CogVLM is supposed to support this with prompts like "Can you provide a description of the image and include the coordinates [[x0,y0,x1,y1]] for each mentioned object?"

However I couldn't get it to work properly, it would just hallucinate.

If you want to give it a shot here are the official visual QA prompts

[–] Scary-Knowledgable@alien.top 1 points 10 months ago

https://github.com/QwenLM/Qwen-VL

[–] takezo07@alien.top 1 points 9 months ago

I'm surprised there is not more options....
As there is LLMs for almost everything!