Context-Aware Single-Shot Detector

By Wei Xiang, Dong-Qing Zhang, Vassilis Athitsos, H. H. Yu. 2017

49 citations recorded by the source corpus.

SSD (Single Shot Detector) is one of the state-of-the-art object detection algorithms, and it combines high detection accuracy with real-time speed. However, it is widely recognized that SSD is less accurate in detecting small objects compared to large objects, because it ignores the context from outside the proposal boxes. In this paper, we present CSSD–a shorthand for context-aware single-shot multibox object detector. CSSD is built on top of SSD, with additional layers modeling multi-scale contexts. We describe two variants of CSSD, which differ in their context layers, using dilated convolution layers (DiCSSD) and deconvolution layers (DeCSSD) respectively. The experimental results show that the multi-scale context modeling significantly improves the detection accuracy. In addition, we study the relationship between effective receptive fields (ERFs) and the theoretical receptive fields (TRFs), particularly on a VGGNet. The empirical results further strengthen our conclusion that SSD coupled with context layers achieves better detection results especially for small objects (+3.2%AP@0.5 on MSCOCO compared to the newest SSD), while maintaining comparable runtime performance.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Radiology, Nuclear Medicine and Imaging

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Sanghyun Woo
2. In-So Kweon
3. Alois C. Knoll
4. Guang Chen
5. Zida Song
6. Zhijun Li
7. Kai Chen
8. Wenkai Chen

In Computer Vision and Pattern Recognition

1. Sanghyun Woo
2. In-So Kweon
3. Alois C. Knoll
4. Guang Chen
5. Zida Song
6. Zhijun Li
7. Kai Chen
8. Wenkai Chen

In Radiology, Nuclear Medicine and Imaging

1. Sanghyun Woo
2. In-So Kweon

Citations in the corpus

• StairNet: Top-Down Semantic Aggregation for Accurate One Shot Detection (2017, background): “0 Caffe DiCSSD300 [37] 07+12 VGGNet 78.”
• A Survey of the Four Pillars for Small Object Detection: Multiscale Representation, Contextual Information, Super-Resolution, and Region Proposal (2022, background): “Meanwhile, a special layer called CSSD is designed to integrate multiscale context information [77].”