DASHA: Decentralized Autofocusing System with Hierarchical Agents
This addresses a domain-specific problem for camera systems and object detection applications, offering a novel solution for auto-focusing without reference data.
The paper tackles the problem of object detection performance degradation due to changes in illumination, environmental conditions, or lens focusing by proposing a decentralized hierarchical multi-agent deep reinforcement learning approach for camera control, resulting in significant improvement beyond popular models like YOLO, Faster R-CNN, and Retina.
State-of-the-art object detection models are frequently trained offline using available datasets, such as ImageNet: large and overly diverse data that are unbalanced and hard to cluster semantically. This kind of training drops the object detection performance should the change in illumination, in the environmental conditions (e.g., rain), or in the lens positioning (out-of-focus blur) occur. We propose a decentralized hierarchical multi-agent deep reinforcement learning approach for intelligently controlling the camera and the lens focusing settings, leading to a significant improvement beyond the capacity of the popular detection models (YOLO, Faster R-CNN, and Retina are considered). The algorithm relies on the latent representation of the camera's stream and, thus, it is the first method to allow a completely no-reference tuning of the camera, where the system trains itself to auto-focus itself.