DSSD : Deconvolutional Single Shot Detector

TLDR

The paper proposes adding contextual information to state‑of‑the‑art general object detection. The method fuses a Residual‑101 classifier with SSD, then augments it with deconvolution layers, a feed‑forward module, and a new output module to provide large‑scale context. On PASCAL VOC and COCO, DSSD achieves 81.5 % mAP on VOC2007, 80.0 % on VOC2012, and 33.2 % on COCO, outperforming R‑FCN.

Abstract

The main contribution of this paper is an approach for introducing additional context into state-of-the-art general object detection. To achieve this we first combine a state-of-the-art classifier (Residual-101[14]) with a fast detection framework (SSD[18]). We then augment SSD+Residual-101 with deconvolution layers to introduce additional large-scale context in object detection and improve accuracy, especially for small objects, calling our resulting system DSSD for deconvolutional single shot detector. While these two contributions are easily described at a high-level, a naive implementation does not succeed. Instead we show that carefully adding additional stages of learned transformations, specifically a module for feed-forward connections in deconvolution and a new output module, enables this new approach and forms a potential way forward for further detection research. Results are shown on both PASCAL VOC and COCO detection. Our DSSD with $513 \times 513$ input achieves 81.5% mAP on VOC2007 test, 80.0% mAP on VOC2012 test, and 33.2% mAP on COCO, outperforming a state-of-the-art method R-FCN[3] on each dataset.