Dynamic Zoom-in Network for Fast Object Detection in Large Images
This work addresses efficiency challenges in object detection for large images, which is incremental as it builds on reinforcement learning to optimize region selection.
The paper tackles the problem of reducing computational cost for object detection in high-resolution images by introducing a dynamic zoom-in framework that processes images in a coarse-to-fine manner, achieving over 50% reduction in processed pixels on Caltech Pedestrians and about 70% reduction with over 50% faster detection time on a high-resolution dataset without accuracy loss.
We introduce a generic framework that reduces the computational cost of object detection while retaining accuracy for scenarios where objects with varied sizes appear in high resolution images. Detection progresses in a coarse-to-fine manner, first on a down-sampled version of the image and then on a sequence of higher resolution regions identified as likely to improve the detection accuracy. Built upon reinforcement learning, our approach consists of a model (R-net) that uses coarse detection results to predict the potential accuracy gain for analyzing a region at a higher resolution and another model (Q-net) that sequentially selects regions to zoom in. Experiments on the Caltech Pedestrians dataset show that our approach reduces the number of processed pixels by over 50% without a drop in detection accuracy. The merits of our approach become more significant on a high resolution test set collected from YFCC100M dataset, where our approach maintains high detection performance while reducing the number of processed pixels by about 70% and the detection time by over 50%.