Ant Financial's LingBot platform introduces LingBot-Depth 2.0, a spatial perception model.

Ant Financial's LingBo Technology launched its spatial perception model, LingBot-Depth 2.0, expanding its training data from 3 million to 150 million images. It achieved first place in 12 out of 16 depth completion benchmark tests, halving the depth error in indoor environments. Simultaneously, it open-sourced its visual pedestal model, LingBot-Vision, featuring industry-first pre-training with boundary structures, achieving sub-pixel-level boundary localization with only 160 million images. LingBot-Depth 2.0 has passed Orbbec certification, and the two companies will collaborate to launch an SDK and integrated camera product.