🚨 Towards Surgical Foundation Model 🚨
Thrilled to announce our NeurIPS 2024 Spotlight: PeskaVLP! A breakthrough in surgical video-language pretraining, setting zero-shot records across diverse surgical tasks!
Paper: https://t.co/2SXW5RbVCK
Code: https://t.co/kp1ZVDdgmM
#NeurIPS
So she was first author on his key paper - in fact she's on most of his key papers, yet she didn't get the Nobel Prize, in spite of the fact they could have given 3 awards, not just 2. Explain to me how STEM is not institutionally sexist.
New paper out at Medical Image Analysis!
https://t.co/qjEs7ch5KD
Using only CNN-LSTMs, we beat SOTA on 3 surgical video tasks, outperforming Transformers.
What did we do?
Turns out, the key components are simply:
- end-to-end training
- a backbone without BatchNorm!
🧵1/5
1/2 Our recent work on Surgical #SSL, accepted at Medical Image Analysis, investigates 4 SSL methods in the surgical domain
#SurgicalDataScience#SurgicalAI
Paper: https://t.co/7n62uTJKU5
Code: https://t.co/6vatu1Ff81
Read the paper to learn
- How many different ways there are to compute the F1 score
- How metrics are implemented with "relaxed boundaries" and why this actually does not do what it should
- How recent methods compare to each other
(2/2)
Are you working on surgical phase recognition?
Our report on evaluation metrics for phase recognition could be interesting for you! (1/2)
ArXiv: https://t.co/wwc6mLcyZ4