Falcon-Perception 0.6B Enables Efficient Local Image Segmentation

vanstriendaniel · x · 2026-08-12

A developer shared their experience using Qwen 3.8 Max for vision tasks, noting that much smaller local models can achieve similar results. They introduced Falcon-Perception (0.6B parameters), an open-vocabulary segmentation model that processes text prompts. By using it for labeling and distilling an even smaller model, the author successfully extracted clean masks for every illustration across over 1 million pages of 19th-century books.

Related event: Falcon-Perception 0.6B Beats SAM 3 in Segmentation(2 posts)→

Original post →

More from Models

Models channel →