Just Add $π$! Pose Induced Video Transformers for Understanding Activities of Daily Living

30 Nov 2023  ·  Dominick Reilly, Srijan Das ·

Video transformers have become the de facto standard for human action recognition, yet their exclusive reliance on the RGB modality still limits their adoption in certain domains. One such domain is Activities of Daily Living (ADL), where RGB alone is not sufficient to distinguish between visually similar actions, or actions observed from multiple viewpoints. To facilitate the adoption of video transformers for ADL, we hypothesize that the augmentation of RGB with human pose information, known for its sensitivity to fine-grained motion and multiple viewpoints, is essential. Consequently, we introduce the first Pose Induced Video Transformer: PI-ViT (or $\pi$-ViT), a novel approach that augments the RGB representations learned by video transformers with 2D and 3D pose information. The key elements of $\pi$-ViT are two plug-in modules, 2D Skeleton Induction Module and 3D Skeleton Induction Module, that are responsible for inducing 2D and 3D pose information into the RGB representations. These modules operate by performing pose-aware auxiliary tasks, a design choice that allows $\pi$-ViT to discard the modules during inference. Notably, $\pi$-ViT achieves the state-of-the-art performance on three prominent ADL datasets, encompassing both real-world and large-scale RGB-D datasets, without requiring poses or additional computational overhead at inference.

PDF Abstract

Results from the Paper


 Ranked #1 on Action Classification on Toyota Smarthome dataset (using extra training data)

     Get a GitHub badge
Task Dataset Model Metric Name Metric Value Global Rank Uses Extra
Training Data
Result Benchmark
Action Recognition NTU RGB+D π-ViT (RGB + Pose) Accuracy (CS) 96.3 # 2
Accuracy (CV) 99.0 # 2
Action Recognition NTU RGB+D π-ViT (RGB only) Accuracy (CS) 94.0 # 10
Accuracy (CV) 97.9 # 9
Action Recognition NTU RGB+D 120 π-ViT (RGB + Pose) Accuracy (Cross-Subject) 96.1 # 1
Accuracy (Cross-Setup) 95.1 # 2
Action Recognition NTU RGB+D 120 π-ViT (RGB only) Accuracy (Cross-Subject) 92.9 # 3
Accuracy (Cross-Setup) 91.9 # 6
Action Classification Toyota Smarthome dataset π-ViT CS 72.9 # 1
CV1 55.2 # 1
CV2 64.8 # 2

Methods


No methods listed for this paper. Add relevant methods here