Collaborative Deep Reinforcement Learning for Joint Object Search
This work addresses the challenge of efficiently localizing interacting objects in computer vision, offering an incremental improvement over existing active localization methods.
The paper tackles the problem of joint active search for multiple interacting objects by introducing a collaborative multi-agent deep reinforcement learning algorithm that exploits contextual cues to improve search efficiency. The method improves state-of-the-art active localization models and reveals interpretable co-detection patterns on multiple benchmarks.
We examine the problem of joint top-down active search of multiple objects under interaction, e.g., person riding a bicycle, cups held by the table, etc.. Such objects under interaction often can provide contextual cues to each other to facilitate more efficient search. By treating each detector as an agent, we present the first collaborative multi-agent deep reinforcement learning algorithm to learn the optimal policy for joint active object localization, which effectively exploits such beneficial contextual information. We learn inter-agent communication through cross connections with gates between the Q-networks, which is facilitated by a novel multi-agent deep Q-learning algorithm with joint exploitation sampling. We verify our proposed method on multiple object detection benchmarks. Not only does our model help to improve the performance of state-of-the-art active localization models, it also reveals interesting co-detection patterns that are intuitively interpretable.