Epic-Sounds: A Large-scale Dataset of Actions That Sound
This provides a new benchmark for audio action recognition, addressing a gap in egocentric video analysis, but it is incremental as it builds on existing dataset and model frameworks.
The authors tackled the problem of audio-based action recognition by introducing EPIC-SOUNDS, a large-scale dataset with 78.4k categorized audio segments across 44 classes, and they trained state-of-the-art models to evaluate audio-only and audio-visual methods.
We introduce EPIC-SOUNDS, a large-scale dataset of audio annotations capturing temporal extents and class labels within the audio stream of the egocentric videos. We propose an annotation pipeline where annotators temporally label distinguishable audio segments and describe the action that could have caused this sound. We identify actions that can be discriminated purely from audio, through grouping these free-form descriptions of audio into classes. For actions that involve objects colliding, we collect human annotations of the materials of these objects (e.g. a glass object being placed on a wooden surface), which we verify from video, discarding ambiguities. Overall, EPIC-SOUNDS includes 78.4k categorised segments of audible events and actions, distributed across 44 classes as well as 39.2k non-categorised segments. We train and evaluate state-of-the-art audio recognition and detection models on our dataset, for both audio-only and audio-visual methods. We also conduct analysis on: the temporal overlap between audio events, the temporal and label correlations between audio and visual modalities, the ambiguities in annotating materials from audio-only input, the importance of audio-only labels and the limitations of current models to understand actions that sound.