Apple sponsored the Conference on Computer Vision and Pattern Recognition (CVPR), which took place virtually from June 14 - 19. CVPR is the premier annual international computer vision event.
Accepted Paper: Least squares binary quantization of neural networks
Hadi Pouransari, Michael Tu, Oncel Tuzel
Quantizing weights and activations of deep neural networks results in significant improvement in inference efficiency at the cost of lower accuracy. A source of the accuracy gap between full precision and quantized models is the quantization error. In this work, we focus on the binary quantization, in which values are mapped to -1 and 1. We provide a unified framework to analyze different scaling strategies. Inspired by the pareto-optimality of 2-bits versus 1-bit quantization, we introduce a novel 2-bits quantization with provably least squares error. Our quantization algorithms can be implemented efficiently on the hardware using bitwise operations. We present proofs to show that our proposed methods are optimal, and also provide empirical error analysis. We conduct experiments on the ImageNet dataset and show a reduced accuracy gap when using the proposed least squares quantization algorithms.
Hadi Pouransari presented the accepted paper on June 15.
The CVPR EarthVision workshop aims at fostering collaboration between the computer vision and Earth Observation communities to boost automated interpretation of remotely sensed data and to raise awareness inside the vision community for this highly challenging and quickly evolving field of research with a big impact on human society, economy, industry, and the planet.
Maros Blaha from Apple gave a virtual keynote talk, entitled Towards Large-Scale Remote Sensing of Human Habitats, on June 14.
Apple sponsored the Women in Computer Vision (WiCV) workshop on June 14. WiCV is committed to raising the visibility of female researchers in the computer vision research community. The workshop focused on showcasing the work of both students and senior researchers through talks and poster sessions, and facilitated networking and career advice between researchers in both academia and industry.