Ant Group's Lingbo Tech launches spatial perception model LingBot-Depth 2.0

Ant Group's Lingbo Tech, a embodied intelligence company, released the spatial perception model LingBot-Depth 2.0, trained on 150 million data samples, with upgrades in edge clarity, small object recognition, and long-range depth estimation. Compared to the previous version, training data expanded from 3 million to 150 million, achieving 12 first-place results out of 16 in depth completion benchmarks. In large indoor depth-missing areas, depth error halved, with RMSE dropping from 0.132 to 0.062. The model performs well in challenging scenarios like glass and mirrors, where traditional depth cameras often fail. Lingbo also launched the vision base model LingBot-Vision, trained on only 160 million images, offering depth estimation accuracy better than DINOv3, and stable boundary tracking in videos. LingBot-Depth 2.0 has been certified by Orbbec's depth vision lab.