A MultiPath Network for Object Detection

TL;DR

This paper introduces a MultiPath network that enhances object detection by integrating multi-layer features, multi-resolution context, and improved localization, significantly boosting performance on the challenging COCO dataset.

Contribution

It proposes a novel MultiPath network architecture with skip connections, a foveal structure, and an integral loss function, combined with DeepMask proposals for improved detection and segmentation.

Findings

66% overall improvement over baseline

4x better detection of small objects

Achieved second place in COCO 2015 challenges

Abstract

The recent COCO object detection dataset presents several new challenges for object detection. In particular, it contains objects at a broad range of scales, less prototypical images, and requires more precise localization. To address these challenges, we test three modifications to the standard Fast R-CNN object detector: (1) skip connections that give the detector access to features at multiple network layers, (2) a foveal structure to exploit object context at multiple object resolutions, and (3) an integral loss function and corresponding network adjustment that improve localization. The result of these modifications is that information can flow along multiple paths in our network, including through features from multiple network layers and from multiple object views. We refer to our modified classifier as a "MultiPath" network. We couple our MultiPath network with DeepMask object proposals, which are well suited for localization and small objects, and adapt our pipeline to predict segmentation masks in addition to bounding boxes. The combined system improves results over the baseline Fast R-CNN detector with Selective Search by 66% overall and by 4x on small objects. It placed second in both the COCO 2015 detection and segmentation challenges.