Cascaded Sparse Spatial Bins for Efficient and Effective Generic Object Detection

TL;DR

This paper introduces a fast, efficient object proposal method combining deep spatial features, edge statistics, and a novel sparse SVM approach, achieving high recall and improved detection performance on standard benchmarks.

Contribution

It presents a new object proposal technique that combines spatial bins, deep features, and sparsity for improved efficiency and recall, outperforming existing methods.

Findings

Achieves state-of-the-art recall on Pascal VOC07.

Attains 78% recall with only 100 proposals per image.

Improves mAP of RCNN by 10 points with 50 proposals.

Abstract

A novel efficient method for extraction of object proposals is introduced. Its "objectness" function exploits deep spatial pyramid features, a novel fast-to-compute HoG-based edge statistic and the EdgeBoxes score. The efficiency is achieved by the use of spatial bins in a novel combination with sparsity-inducing group normalized SVM. State-of-the-art recall performance is achieved on Pascal VOC07, significantly outperforming methods with comparable speed. Interestingly, when only 100 proposals per image are considered the method attains 78% recall on VOC07. The method improves mAP of the RCNN state-of-the-art class-specific detector, increasing it by 10 points when only 50 proposals are used in each image. The system trained on twenty classes performs well on the two hundred class ILSVRC2013 set confirming generalization capability.