Modular Action Concept Grounding in Semantic Video Prediction
This work addresses the problem of learning agent-object interactions for video prediction, which is crucial for developing higher-level cognitive abilities in AI systems.
This paper introduces semantic action-conditional video prediction, a task that uses semantic action labels to describe interactions between agents and objects. The proposed Modular Action Concept Network (MAC) effectively conditions on given instructions and generates corresponding future frames, demonstrating out-of-distribution generalization and quick adaptation to new object categories.
Recent works in video prediction have mainly focused on passive forecasting and low-level action-conditional prediction, which sidesteps the learning of interaction between agents and objects. We introduce the task of semantic action-conditional video prediction, which uses semantic action labels to describe those interactions and can be regarded as an inverse problem of action recognition. The challenge of this new task primarily lies in how to effectively inform the model of semantic action information. Inspired by the idea of Mixture of Experts, we embody each abstract label by a structured combination of various visual concept learners and propose a novel video prediction model, Modular Action Concept Network (MAC). Our method is evaluated on two newly designed synthetic datasets, CLEVR-Building-Blocks and Sapien-Kitchen, and one real-world dataset called Tower-Creation. Extensive experiments demonstrate that MAC can correctly condition on given instructions and generate corresponding future frames without need of bounding boxes. We further show that the trained model can make out-of-distribution generalization, be quickly adapted to new object categories and exploit its learnt features for object detection, showing the progression towards higher-level cognitive abilities. More visualizations can be found at http://www.pair.toronto.edu/mac/.