Zoom Out-and-In Network with Recursive Training for Object Proposal

TL;DR

This paper introduces a multi-resolution zoom out-and-in network with recursive training for improved object proposal generation, effectively detecting objects of various sizes and enhancing detection accuracy.

Contribution

It presents a novel zoom-out-and-in architecture with recursive training, combining high and low-level features for better object proposals across scales.

Findings

Outperforms state-of-the-art methods on ILSVRC DET and MS COCO datasets.

Increases average precision by around 2% in detection systems.

Effectively detects small, medium, and large objects using multi-scale feature maps.

Abstract

In this paper, we propose a zoom-out-and-in network for generating object proposals. We utilize different resolutions of feature maps in the network to detect object instances of various sizes. Specifically, we divide the anchor candidates into three clusters based on the scale size and place them on feature maps of distinct strides to detect small, medium and large objects, respectively. Deeper feature maps contain region-level semantics which can help shallow counterparts to identify small objects. Therefore we design a zoom-in sub-network to increase the resolution of high level features via a deconvolution operation. The high-level features with high resolution are then combined and merged with low-level features to detect objects. Furthermore, we devise a recursive training pipeline to consecutively regress region proposals at the training stage in order to match the iterative regression at the testing stage. We demonstrate the effectiveness of the proposed method on ILSVRC DET and MS COCO datasets, where our algorithm performs better than the state-of-the-arts in various evaluation metrics. It also increases average precision by around 2% in the detection system.