Ashish Patel 🇮🇳’s Post

𝗗𝗮𝘆-𝟭𝟴𝟭 Computer Vision Learning 𝗣𝗼𝗹𝗸𝗮 𝗟𝗶𝗻𝗲𝘀: Learning Structured Illumination and Reconstruction for Active Stereo by Princeton University Follow me for a similar post:  🇮🇳 Ashish Patel Interesting Facts : 🔸 This is a paper in CVPR 2021 with over 3 citations. ------------------------------------------------------------------- 𝗔𝗺𝗮𝘇𝗶𝗻𝗴 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 : https://lnkd.in/eammrcV ------------------------------------------------------------------- 𝗜𝗠𝗣𝗢𝗥𝗧𝗔𝗡𝗖𝗘 🔸 Active stereo cameras that recover depth from structured light captures have become a cornerstone sensor modality for 3D scene reconstruction and understanding tasks across application domains. Active stereo cameras project a pseudo-random dot pattern on object surfaces to extract disparity independently of object texture.  🔸 A novel method for learning an active stereo camera, including illumination, capture, and depth reconstruction. Departing from hand-engineered illumination patterns, it learns novel illumination patterns, the Polka Lines patterns, that provide state-of-the-art depth reconstruction and insights on the function of structured illumination patterns under various imaging conditions.  🔸 To realize this approach, Introduce a hybrid image formation model that exploits both wave optics and geometric optics for efficient end-to-end optimization, and a trinocular reconstruction network that exploits the trinocular depth cues of active stereo systems. 🔸 The proposed method allows us to design environment-specific structured Polka Line patterns tailored to the camera and scene statistics. We validate the effectiveness of our approach with comprehensive simulations and with an experimental prototype, outperforming conventional hand-crafted patterns across all tested scenarios. #computervision #artificialintelligence #data