We propose a self-supervised algorithm to learn representations from egocentric video data. Recently, significant efforts have been made to capture humans interacting with their own environments as they go about their daily activities. In result, several large egocentric datasets of interaction-rich multi-modal data have emerged. However, learning representations from videos can be challenging. First, given the uncurated nature of long-form continuous videos, learning effectiverepresentations require focusing onmoments intimewhen interactions take place. Second, visual representations of daily activities should be sensitive to changes in the state of the environment. However, current successful multimodal learning frameworks encourage representation invariance over time.

While these approaches learn high-quality representations for downstream tasks such as action recognition, their training objectives disregard spatial cues naturally occurring in audio and visual signals.

The authors propose a vision method to estimate where within an image a pictured head is looking. Given the image and the cropped out head, the system returns a saliency map consisting of confidence ratings on grid cells saying how likely that position is to be the subject of that head (person)'s gaze. The technique uses CNN with two pathways, one for the head/gaze and one for the full image/saliency of the scene. This dataset contains in total 36K people. The method is compared to a few reasonable baselines that represent alternative approaches one might implement and sanity checks.

An environment acoustic model represents how sound is transformed by the physical characteristics of an indoor environment, for any given source/receiver location. Traditional methods for constructing acoustic models involve expensive and time-consuming collection of large quantities of acoustic data at dense spatial locations in the space, or rely on privileged knowledge of scene geometry to intelligently select acoustic data sampling locations. We propose active acoustic sampling, a new task for efficiently building an environment acoustic model of an unmapped environment in which a mobile agent equipped with visual and acoustic sensors jointly constructs the environment acoustic model and the occupancy map on-the-fly. We introduce ActiveRIR, a reinforcement learning (RL) policy that leverages information from audio-visual sensor streams to guide agent navigation and determine optimal acoustic data sampling positions, yielding a high quality acoustic model of the environment from a minimal set of acoustic samples. We train our policy with a novel RL reward based on information gain in the environment acoustic model. Evaluating on diverse unseen indoor environments from a state-of-the-art acoustic simulation platform, ActiveRIR outperforms an array of methods--both traditional navigation agents based on spatial novelty and visual exploration as well as existing state-of-the-art methods.

Developing embodied agents in simulation has been a key research topic in recent years. Exciting new tasks, algorithms, and benchmarks have been developed in various simulators. However, most of them assume deaf agents in silent environments, while we humans perceive the world with multiple senses. We introduce Sonicverse, a multisensory simulation platform with integrated audio-visual simulation for training household agents that can both see and hear. Sonicverse models realistic continuous audio rendering in 3D environments in real-time. Together with a new audio-visual VR interface that allows humans to interact with agents with audio, Sonicverse enables a series of embodied AI tasks that need audio-visual perception. For semantic audio-visual navigation in particular, we also propose a new multi-task learning model that achieves state-of-the-art performance. In addition, we demonstrate Sonicverse's realism via sim-to-real transfer, which has not been achieved by other simulators: an agent trained in Sonicverse can successfully perform audio-visual navigation in real-world environments. Sonicverse is available at: https://github.com/StanfordVL/Sonicverse.

Facebook announced a research project Thursday that aims to develop an artificial intelligence capable of perceiving the world like a human being. The project, titled Ego4D, aims to train an artificial intelligence (AI) to perceive the world in the first-person by analyzing a constant stream of video from people's lives. This type of data, which Facebook calls "egocentric" data, is designed to help the AI perceive, remember and plan like a human being. "Next-generation AI systems will need to learn from an entirely different kind of data -- videos that show the world from the center of the action, rather than the sidelines," Kristen Grauman, lead AI research scientist at Facebook, said in the announcement. The project aims to improve AI technology's capacity to accomplish human processes by setting five key benchmarks: "episodic memory," in which the AI ties memories to specific locations and times, "forecasting," "social interaction," "hand and object manipulation" and "audio-visual diarization," in which the AI ties auditory experiences to specific locations and times.

AI that understands the world from a first-person point of view could unlock a new era of immersive experiences, as devices like augmented reality (AR) glasses and virtual reality (VR) headsets become as useful in everyday life as smartphones. Imagine your AR device displaying exactly how to hold the sticks during a drum lesson, guiding you through a recipe, helping you find your lost keys, or recalling memories as holograms that come to life in front of you. To build these new technologies, we need to teach AI to understand and interact with the world like we do, from a first-person perspective -- commonly referred to in the research community as egocentric perception. Today's computer vision (CV) systems, however, typically learn from millions of photos and videos that are captured in third-person perspective, where the camera is just a spectator to the action. "Next-generation AI systems will need to learn from an entirely different kind of data -- videos that show the world from the center of the action, rather than the sidelines," says Kristen Grauman, lead research scientist at Facebook.

To operate in augmented and virtual reality, Facebook believes artificial intelligence will need to develop an "egocentric perspective." To that end, the company on Thursday announced Ego4D, a data set of 2,792 hours of first-person video, and a set of benchmark tests for neural nets, designed to encourage the development of AI that is savvier about what it's like to move through virtual worlds from a first-person perspective. The project is a collaboration between Facebook Reality Labs and scholars from 13 research institutions, including academic institutions and research labs. The details are laid out in a paper lead-authored by Facebook's Kristen Grauman, "Ego4D: Around the World in 2.8K Hours of Egocentric Video." Grauman is a scientist with the company's Facebook AI Research unit.

For the last two years, Facebook AI Research (FAIR) has worked with 13 universities around the world to assemble the largest ever data set of first-person video--specifically to train deep-learning image-recognition models. AIs trained on the data set will be better at controlling robots that interact with people, or interpreting images from smart glasses. "Machines will be able to help us in our daily lives only if they really understand the world through our eyes," says Kristen Grauman at FAIR, who leads the project. Such tech could support people who need assistance around the home, or guide people in tasks they are learning to complete. "The video in this data set is much closer to how humans observe the world," says Michael Ryoo, a computer vision researcher at Google Brain and Stony Brook University in New York, who is not involved in Ego4D.

Facebook is pouring a lot of time and money into augmented reality, including building its own AR glasses with Ray-Ban. Right now, these gadgets can only record and share imagery, but what does the company think such devices will be used for in the future? A new research project led by Facebook's AI team suggests the scope of the company's ambitions. It imagines AI systems that are constantly analyzing peoples' lives using first-person video; recording what they see, do, and hear in order to help them with everyday tasks. Facebook's researchers have outlined a series of skills it wants these systems to develop, including "episodic memory" (answering questions like "where did I leave my keys?") and "audio-visual diarization" (remembering who said what when).