In the last decade, there has been a significant advancement in capturing and rendering 3D Computer Vision. However, there is very little awareness about it amidst the community, and only limited groups in India work in this domain. The purpose of the Summer School was to bring exposure to the various research problems and existing algorithms in the domain of 3D Computer Vision, typically seen as a subset of computer vision. “The idea was to introduce and give exposure to students, professionals, and industrialists to 3D Vision problems since it’s not very popular and not many people are doing it in India,” said Dr. Charu Sharma. A 3D Summer School was initially conducted in February of 2020 as a small-scale event and this year a second version was organized. The Summer School covered 3D Vision fundamentals by invited speakers from across the world as well as by the faculty and a Ph.D. student from CVIT, IIITH.
The 82 attendees from across India comprised 31 industrialists, 51 students( 3 rd and 4 th year), academicians and researchers from IIITs, NITs, IITs as well as other top colleges. Minimal criteria were followed to select the attendees: whether there is prior exposure to the domain; if the relevant data representation and analysis techniques were used in academic or industrial research; and if any prerequisites are completed in image processing and deep learning in the degree the students are currently pursuing.
The Summer School began with an “Introduction to Image Processing”, followed by “Primer on 2D Deep Learning Neural Networks, MLPs, CNNs” by Dr. Ravi Kiran, Professor from CVIT at IIITH. Essentially, a wide range of 3D object representations, reconstruction and analysis techniques were discussed. The second day undertook with a “Welcome Address” by Prof. P. J. Narayanan, Director of IIITH. The invited speakers from across the world and the topics that were covered during the week-long event were as follows:
- Prof. Madhava Krishna (Professor at IIITH) - "The SLAM Pipeline and Beyond"
- Dr. Arjun Jain (Founder and Chief Scientist at UAVIO Labs Pvt. Ltd.) - "Self-Driving at scale: Where are we headed" and "3D humans: From stick figures to digital avatars"
- Dr. Varun Jampani( Researcher at Google Research) - "Practical 3D Object Understanding from Image Collections and Videos"
- Dr. Abhijeet Kundu (Researcher at Google Research) - "Panoptic Neural Fields: A Semantic Object-Aware Neural Scene Representation"
- Dr. Srinath Sridhar(Assistant Professor at Brown University) - "Canonicalization for Object-Centric 3D Scene Understanding".
Assistant Professors - a) Dr. Avinash Sharma, b) Dr. Charu Sharma and c) Dr. Ravi Kiran
The following sessions were also conducted by the faculty of IIITH:
- “3D Sensing, Representation and Visualization”, “Introduction to Geometry Processing” and “Introduction to Implicit Function Learning” by Dr. Avinash Sharma
- “Introduction to 3D Deep Volumetric & Point Cloud Learning (Part 1 and 2)” and “Introduction to Deep Learning on Meshes” by Dr. Charu Sharma
The session, 3D Human Reconstruction, was handled by Jinka Sai Sagar, a Ph.D. student from CVIT, IIITH. The daily hands-on sessions handled by the volunteers were:
- Image Processing, MLP, CNN
- Blender, Kinect Sensing
- Open3D, ICP, MeshLab, Trimesh
- VoxelNet, Pointnet
- MeshCNN, GCN
- Implicit Function Learning
- NERF
The 3D Summer School concluded on Friday, May 28, 2022, with a Valedictory Session followed by a Banquet Dinner. “The intention of this 3D Vision Summer School was to bring exposure to the various academic and industrial groups across the country, so people are enabled to contribute better towards the ecosystem of 3D Computer Vision”, commented Prof. Avinash Sharma. As interesting ideas were exchanged, regular reading groups are intended to be organized across the country as a follow-up. Following the success of the 3D Vision Summer School, the team hopes to conduct a more dynamic event next year!