Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
enoch2090
3 months ago
|
parent
|
context
|
favorite
| on:
Meta Segment Anything Model 3
Surprisingly, SAM3 works bad on engineering drawings while SAM2 kinda works, and VLMs like Qwen3-VL works as well
zubiaur
3 months ago
|
next
[–]
Had good luck with Gemini 2.5, SAM3 failed miserably with PIDs.
retinaros
3 months ago
|
prev
[–]
yeah I tried too. Im trying a fine tuning on PIDs.
enoch2090
3 months ago
|
parent
[–]
Looking forward to your progress! Just checked the paper and it says the underlying backbone is still DETR. My guess would be that SAM3 uses more video frames during the training process and caused the dilution of sparse engineering-paper-like data.
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
