Streaming egocentric action anticipation: An evaluation scheme and approach

TL;DR

This paper introduces a streaming evaluation scheme for egocentric action anticipation that accounts for model runtime, and proposes a lightweight model optimized with knowledge distillation, leading to more realistic performance assessments.

Contribution

It presents a new streaming evaluation framework and a lightweight 3D CNN model with knowledge distillation for egocentric action anticipation.

Findings

Streaming evaluation alters model rankings compared to traditional methods.

Lightweight models outperform more complex ones in streaming scenarios.

The proposed model surpasses current state-of-the-art in streaming egocentric action anticipation.

Abstract

Egocentric action anticipation aims to predict the future actions the camera wearer will perform from the observation of the past. While predictions about the future should be available before the predicted events take place, most approaches do not pay attention to the computational time required to make such predictions. As a result, current evaluation schemes assume that predictions are available right after the input video is observed, i.e., presuming a negligible runtime, which may lead to overly optimistic evaluations. We propose a streaming egocentric action evaluation scheme which assumes that predictions are performed online and made available only after the model has processed the current input segment, which depends on its runtime. To evaluate all models considering the same prediction horizon, we hence propose that slower models should base their predictions on temporal segments sampled ahead of time. Based on the observation that model runtime can affect performance in the considered streaming evaluation scenario, we further propose a lightweight action anticipation model based on feed-forward 3D CNNs which is optimized using knowledge distillation techniques with a novel past-to-future distillation loss. Experiments on the three popular datasets EPIC-KITCHENS-55, EPIC-KITCHENS-100 and EGTEA Gaze+ show that (i) the proposed evaluation scheme induces a different ranking on state-of-the-art methods as compared to classic evaluations, (ii) lightweight approaches tend to outmatch more computationally expensive ones, and (iii) the proposed model based on feed-forward 3D CNNs and knowledge distillation outperforms current art in the streaming egocentric action anticipation scenario.