Sunando Sengupta

Sunando Sengupta is a computer-vision researcher in Microsoft’s Applied Sciences Group (ASG).  His research interests include semantic scene understanding, 3D scene reconstruction, optimizing vision models for hardware, and vision for health care, among others. 

He did his Ph.D. work in computer vision with Philipp Torr at Oxford Brookes University, where he worked on road-scene classification and reconstruction.

After earning his Ph.D., Sunando worked in multiple industry sectors.  He did motion-capture engineering for the entertainment industry; designed camera-based monitors to track vital signs such as breathing and heart rates in the health-care field.  Most recently before coming to ASG, he spent two years at Intel working on such projects as sensor fusion for autonomous driving systems and low-precision deep-learning models to target field-programmable gate arrays (FPGA).  His work has been presented at various industrial trade shows.

In his free time Sunando likes to ride his bike and drink real ales.

• Adaptive Token Sampling For Efficient Vision Transformers

  European Conference on Computer Vision (ECCV) 2022 July, 2022 Vol. abs/2111.15667v3

  By Mohsen Fayyaz, Soroush Abbasi Koohpayegani, Farnoush Rezaei Jafari, Sunando Sengupta, Hamid Vaezi Joze, Eric Sommerlade, Hamed Pirsiavash, Juergen Gall