Publications

  1. Make3D: Learning 3-D Scene Structure from a Single Still Image, Ashutosh Saxena, Min Sun, Andrew Y. Ng, To appear in IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2008. [pdf]

    (Journal version of ICCV-3dRR'07 and ICCV-VRML'07 with additional details on the algorithm and the learning/inference algorithm.)

  2. Make3D: Depth Perception from a Single Still Image, Ashutosh Saxena, Min Sun, Andrew Y. Ng, In AAAI (Nectar track), 2008. [pdf]

    (This paper provides intuition on the general idea of depth perception, and its applications to various fields. Giving a high-level description (i.e., no mathematics), this paper summarizes our papers from 2005 till 2007.)

  3. Learning 3-D Scene Structure from a Single Still Image, Ashutosh Saxena, Min Sun, Andrew Y. Ng, In ICCV workshop on 3D Representation for Recognition (3dRR-07), 2007. (Best paper award.) [pdf, ppt, Talk]

    (State of the art algorithm in Nov 2006, which produces quantitatively accurate 3-d depths as well as visually pleasing fully textured 3-d flythroughs from an image. It proposes learning a novel parameterization suited to scene understanding. More.)

  4. 3-D Reconstruction from Sparse Views using Monocular Vision, Ashutosh Saxena, Min Sun, Andrew Y. Ng, In ICCV workshop on Virtual Representations and Modeling of Large-scale environments (VRML), 2007. [pdf]

    (Algorithm to build large scale 3d models from a few images, even with little or no overlap. More.)

  5. 3-D Depth Reconstruction from a Single Still Image, Ashutosh Saxena, Sung H. Chung, Andrew Y. Ng. International Journal of Computer Vision (IJCV), Aug 2007. [pdf]

    (Journal version of NIPS 2005 and IJCAI 2007 papers. More.)

  6. Depth Estimation using Monocular and Stereo Cues, Ashutosh Saxena, Jamie Schulte, Andrew Y. Ng. In International Joint Conference on Artificial Intelligence (IJCAI), 2007. [pdf]

    (Monocular cues were used to improve the performance of stereo vision. More.)

  7. High Speed Obstacle Avoidance using Monocular Vision and Reinforcement Learning, Jeff Michels, Ashutosh Saxena, Andrew Y. Ng. In International Conference on Machine Learning (ICML), 2005. [pdf, ppt]

    (A simplified version of the algorithm was used to drive a rccar autonomously in realtime. More.)

  8. Learning Depth from Single Monocular Images, Ashutosh Saxena, Sung H. Chung, Andrew Y. Ng. In Neural Information Processing Systems (NIPS) 18, 2005. [pdf]

    (Infer a depthmap from a single still image. More.)

Others

  1. Rapid Interactive 3D Reconstruction from a Single Still Image, Ashutosh Saxena, Nuwan Senaratna, Savil Srivastava, Andrew Y. Ng. In SIGGRAPH Late Breaking work (Informal Session), 2008. [email us for pdf]

    (Create 3D models/flythroughs using user scribbles---just a few seconds of user input.)

  2. Building a 3-D Model from a Single Still Image, Ashutosh Saxena, Min Sun and Andrew Y. Ng, Demonstration in Neural Information Processing Systems (NIPS), 2007.

    Also presented at NIPS Workshop on The Grammar of Vision: Probabilistic Grammar-Based Models for Visual Scene Understanding and Object Categorization, 2007. [png]

    To be also presented at AAAI Demonstration, 2008.

    (Showing the beta version of the technology and the website. More.)

 

Suggested readings: Papers number 3 (ICCV-3dRR 2007), 5 (IJCV 2007) and 8 (NIPS 2005) give a good coverage of the monocular vision work. For a more informal summary, read 2 (AAAI-nectar).

 

Previous Results

  1. See Nov 2006 models (588 from internet images in ICCV-3dRR, and large-scale ones in ICCV-VRML) here.

    (For viewing VRMLs, we suggest Cortona plugin for Windows, and this for Linux. Takes less than a minute to install.)

  2. See NIPS 2005 results here.

 

Digg it | Reddit it | Slashdot