@ModelScope2022: AMAP CV Lab introduces ABot-Recon, turning long video streams into camera paths and 3D point clouds in real time. https…
Summary
AMAP CV Lab has introduced ABot-Recon, a real-time AI system that processes long video streams to generate 3D point clouds and camera paths, featuring stable performance and open-source code released under Apache 2.0.
View Cached Full Text
Cached at: 08/28/26, 05:45 AM
AMAP CV Lab introduces ABot-Recon, turning long video streams into camera paths and 3D point clouds in real time. https://modelscope.ai/models/amap_cvlab/ABot-Recon…
4.35 m ATE and 91.81% F1 on Oxford Spires, without loop closure. 24.45 FPS with only 6.71 GiB memory in the official test. Looks at just the latest 12 frames, so speed and memory stay stable even as videos get longer. Supports streams up to 22,000 frames and builds them into one continuous 3D reconstruction. Checkpoint, inference, and evaluation code released. Code: Apache 2.0; weights use separate model terms.
Similar Articles
Revisiting Local Context for Long-Horizon Streaming 3D Reconstruction
ABot-Recon enables real-time 3D reconstruction of large-scale environments from continuous video streams by using a fixed local context, achieving high efficiency and low memory usage.
@IlirAliu_: Forget lidar. One single camera. Runs in real time & is open source: A streaming 3D model that reconstructs scenes live…
LingBot-Map is an open-source, real-time streaming 3D reconstruction model that uses a single camera, running at ~20 FPS via a feed-forward geometric context transformer, outperforming both streaming and offline methods.
robbyant/lingbot-map
LingBot-Map is a feed-forward 3D foundation model for streaming 3D reconstruction that uses a Geometric Context Transformer architecture, achieving state-of-the-art performance with efficient ~20 FPS inference on long sequences exceeding 10,000 frames.
SAM 3.1: Faster and More Accessible Real-Time Video Detection and Tracking With Multiplexing and Global Reasoning
Meta AI releases SAM 3.1, an update to the Segment Anything Model that enhances real-time video detection and tracking through multiplexing and global reasoning capabilities.
@skalskip92: there's no catch; SAM3 is open source and really good one of the things it does really well is object tracking, even in…
SAM3 (Segment Anything Model 3) is open source and performs exceptionally well at object tracking even in complex scenes like basketball, making it a standout computer vision model.