Fast Object Detection with Latticed Multi-Scale Feature Fusion

TL;DR

This paper introduces FluffNet, a real-time multi-scale object detector using the novel Fluff block for efficient, fine-grained feature fusion, achieving state-of-the-art accuracy on MS COCO and PASCAL VOC datasets.

Contribution

The paper proposes the Fluff block, a new multi-scale feature fusion module that improves detection accuracy and efficiency, and demonstrates its integration into SSD as FluffNet.

Findings

Achieves state-of-the-art accuracy on MS COCO and PASCAL VOC datasets.

Provides a highly efficient, real-time multi-scale object detection method.

Shows the Fluff block's versatility when embedded into other detectors.

Abstract

Scale variance is one of the crucial challenges in multi-scale object detection. Early approaches address this problem by exploiting the image and feature pyramid, which raises suboptimal results with computation burden and constrains from inherent network structures. Pioneering works also propose multi-scale (i.e., multi-level and multi-branch) feature fusions to remedy the issue and have achieved encouraging progress. However, existing fusions still have certain limitations such as feature scale inconsistency, ignorance of level-wise semantic transformation, and coarse granularity. In this work, we present a novel module, the Fluff block, to alleviate drawbacks of current multi-scale fusion methods and facilitate multi-scale object detection. Specifically, Fluff leverages both multi-level and multi-branch schemes with dilated convolutions to have rapid, effective and finer-grained feature fusions. Furthermore, we integrate Fluff to SSD as FluffNet, a powerful real-time single-stage detector for multi-scale object detection. Empirical results on MS COCO and PASCAL VOC have demonstrated that FluffNet obtains remarkable efficiency with state-of-the-art accuracy. Additionally, we indicate the great generality of the Fluff block by showing how to embed it to other widely-used detectors as well.