Although we've seen a recent boost of sensors in digital devices, e.g., virtual reality headsets, autonomous driving, smart home, IoT devices, advanced algorithms for processing and handling visual data in the backend are largely missing. One key challenge is that algorithms that can understand images, human behaviors, and social activities from sensors deployed in daily lives go beyond the traditional scope of image and scene understanding, and they are expected to be capable of answering queries much broader and deeper than “what is where”. The mission of this workshop is to (a) identify the key domains in modern computer vision; (b) formalize the computational challenges in these domains; and (c) provide promising frameworks to solve these challenges.