Interns at InterDigital: Exploring Vision for Machines
Throughout 2023, InterDigital’s New York Office welcomed more than a dozen brilliant interns and researchers to explore and contribute to the industry-shaping research and innovation led at InterDigital.
Before concluding their time at InterDigital, each of the interns reflected on the work they led and experiences they shared working in InterDigital’s esteemed labs with world class engineers and inventors.
We thank each of our interns for their insights, their diligence, and the hard work they contributed to the Lab’s research and our InterDigital team. Learn more about the interns and their work below.
Eric Lei: Leveraging a Novel Mesh Autoencoder
Philip Jacobson: Generating Point Cloud Texture using Generative AI
Ritwik Sadhu: Exploring Geometry Regularized Point Cloud Autoencoder
Leveraging a Novel Mesh Autoencoder
Throughout his internship, Eric Lei worked on 3D mesh learning and compression. Eric’s project first involved designing a novel mesh autoencoder that allows autoencoding of meshes that do not share template graphs, increasing the applicability of autoencoder tasks in the 3D graphics domain. His follow up work includes adapting the autoencoder for mesh compression, providing a neural compression approach for mesh data.
Eric Lei: “During my time at InterDigital, I was able to explore many timely research topics in 3D graphics learning and compression. I learned a lot about 3D graphics data and was able to merge what I learned with my prior experience in learned compression and compression theory to develop interesting solutions. I thoroughly enjoyed discussions with Muhammad Asad Lodhi, Dong Tian, Junghyun Ahn, and Jiahao Pang, who all possess exceptional expertise in this field and provided me with great support. Overall, I had a pleasure working with the V4M team and thank them for my time at InterDigital.”
Generating Point Cloud Texture using Generative AI
At InterDigital, Philip Jacobson explored the topic of point cloud color and attribute generation using generative AI. Given a 3D point cloud and a text description, InterDigital’s developed approach colorizes the input point cloud as instructed. With a diffusion model operating on 3D point clouds, our network iteratively improves the colorization and leads to the desired output. In this 3-month period, Philip proposed practical solutions and demonstrated his strong knowledge in this subject, leading to potential patent filing and publication.
Philip Jacobson: “I interned at InterDigital for 3 months in summer 2023 working on a team dedicated to researching learning-based algorithms for point cloud analysis and processing. For my intern project, I explored using generative AI to create realistic textures for point clouds, which could be used in AR/VR/gaming and facilitated the development of point cloud attribute compression. I enjoyed being able to explore something outside of my current PhD research track in an exciting, fast-moving direction. The mentorship I received from my mentor Jiahao, and the talented and knowledgeable team has helped me grow tremendously as a researcher and will stick with me as I look to advance my career.”
Exploring Geometry Regularized Point Cloud Autoencoder
While at InterDigital, Ritwik Sadhu explored the topic of 3D point cloud surface analysis, leading to a loss function that greatly improves the point cloud reconstruction quality and feature representability in point-based neural network training. During his internship, Ritwik successfully synergized his expertise in optimal transport with 3D point cloud analysis and shows his enthusiasm for the field. His work will be integrated into InterDigital’s AI-based point cloud compression project for additional improvement of the compression performance.
Ritwik Sadhu: “I have interned at InterDigital for the past 3 months, working on 3D point cloud surface analysis for measuring the similarity of 3D point clouds representing real-life objects. Starting as a machine learning generalist and ending with a potential patent and research publication on my solution, this internship has been a tour-de-force in understanding the state-of-the-art in the field of point cloud processing, and the support from Jiahao Pang and Dong Tian was phenomenal. I deeply enjoyed the brainstorming sessions with the V4M team, which helped refine my ideas and experiments – I thank all of them for their help and company this summer.”