Salman Khan (@KhanSalmanH)

2025-01-23 | โค๏ธ 177 | ๐Ÿ” 41


๐Ÿ”ฅ Open-YOLO 3D is accepted to ICLR! Congratulations to Mohamed El Amine and all the amazing team members ๐Ÿ‘

TLDR: itโ€™s a SoTA model on open-vocabulary 3D instance segmentation with up to 16x speedup. Checkout our project at: https://github.com/aminebdj/OpenYOLO3D

๐Ÿ”— ์›๋ณธ ๋งํฌ


Auto-generated - needs manual review

์ธ์šฉ ํŠธ์œ—

Salman Khan (@KhanSalmanH)

๐Ÿš€Recent open-vocabulary 3D instance segmentation methods struggle with slow inference speed. Our new approach, Open-YOLO 3D, dramatically accelerates the speed up to 16x while maintaining high accuracy.

Paper: https://t.co/Oh4p0u2D5P Code: https://t.co/Q4OfqcM0N1 https://t.co/Dy6QfAFAqz

์›๋ณธ ํŠธ์œ—

quoted-image

Tags

domain-vision-3d domain-visionos