Note: This video may require joining the NVIDIA Developer Program or login

GTC Silicon Valley-2019 ID:S9599:Towards Weakly Supervised Scene Understanding with Deep Latent Variable Models and Structured Priors

Zhiding Yu(NVIDIA)
Our talk will cover the problem of weakly supervised learning for visual scene understanding. We'll explain the background for our research, the challenges involved, and why this research is important. We'll also touch on related work. Our talk will examine how we're addressing challenges and highlight two recent ECCV18 papers on edge alignment/unsupervised domain adaptation. We'll describe our ongoing work on weakly supervised object detection with image-level labels and look into some future directions for research.

View the slides (pdf)