SenseTime open-sources SenseNova-Vision unified vision model

SenseTime has fully open-sourced SenseNova-Vision, a unified multimodal vision model.

The model handles detection, OCR, segmentation, depth estimation, 3D reconstruction, surface-normal prediction, and multi-view geometry.

SenseTime also released the SenseNova-Vision Corpus-50M training dataset and plans to integrate the model into the SenseNova U-series.

The move could reduce reliance on separate task-specific computer vision systems and accelerate adoption of multimodal AI tools.