Public/automatic
1
0
Fork 0
mirror of https://github.com/vladmandic/automatic synced 2026-09-20 01:31:13 +02:00
Code Issues Packages Projects Releases Wiki Activity

fix(vqa): handle Moondream point and detect tasks

Browse Source 
Add handlers for "Point at..." and "Detect..." tasks in moondream()
that were falling through to answer_question() and failing.
This commit is contained in:
CalamitousFelicitousness
2025-12-04 21:48:06 +00:00
parent 506515b018
commit c75a09be83
1 changed files with 8 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files
modules/interrogate/vqa.py
+8 -3
View File
@@ -1111,11 +1111,16 @@ def moondream(question: str, image: Image.Image, repo: str = None, model_name: s
response = model.caption(image, length="normal")['caption']
elif question == 'MORE DETAILED CAPTION':
response = model.caption(image, length="long")['caption']
elif question.lower().startswith('point at '):
target = question[9:]
result = model.point(image, target)
response = str(result)
elif question.lower().startswith('detect '):
target = question[7:]
result = model.detect(image, target)
response = str(result)
else:
response = model.answer_question(encoded, question, processor)['answer']
# model.detect(image, "face")
# model.point(image, "person")
# model.detect_gaze(image)
return response