CVPR Daily - Tuesday

2 Francesca's Picks DAILY T u e s d a y For Today, Tuesday 16 th Francesca Babiloni is currently working as a Computer Vision Research Scientist at Huawei's Noah's Ark Lab in London. “Our lab focuses on computational photography and 3D vision for smartphone applications, with a sweet spot for innovation in deep learning architectures and systems.” “I am working on real datasets as well as attention and non-local blocks in neural networks. I am very happy! We get the chance to do research and have impact on technology utilized by millions.” “It's sad that CVPR is going to be virtual this year, especially for the chat at the poster sessions. Not to mention that everybody is going to have a gap in their CVPR t-shirt collection now!” Francesca should have mentioned her own poster! She is presenting TESA: Tensor Element Self-Attention via Matricization on Thursday at 5:00 and 17:00. Don't miss it! Francesca's picks of the day: Orals: 1.2B 15 Deep White-Balance Editing 1.2C 22 AdderNet: Do We Really Need Multiplications in Deep Learning? 1.4A 3 Dynamic Graph Message Passing Networks Posters: 1.2 64 Self2Self With Dropout: Learning Self-Supervised Denoising From Single Image 1.2 107 Dynamic Convolutions: Exploiting Spatial Sparsity for Faster Inference 1.3 61 From Fidelity to Perceptual Quality: A Semi-Supervised Approach for Low-Light Image Enhancement