Single-Shot Object Detection with Enriched Semantics

By Zhishuai Zhang, Siyuan Qiao, Cihang Xie, Wei Shen, Alan L. Yuille. 2017

218 citations recorded by the source corpus.

We propose a novel single shot object detection network named Detection with Enriched Semantics (DES). Our motivation is to enrich the semantics of object detection features within a typical deep detector, by a semantic segmentation branch and a global activation module. The segmentation branch is supervised by weak segmentation ground-truth, i.e., no extra annotation is required. In conjunction with that, we employ a global activation module which learns relationship between channels and object classes in a self-supervised manner. Comprehensive experimental results on both PASCAL VOC and MS COCO detection datasets demonstrate the effectiveness of the proposed method. In particular, with a VGG16 based DES, we achieve an mAP of 81.7 on VOC2007 test and an mAP of 32.8 on COCO test-dev with an inference speed of 31.5 milliseconds per image on a Titan Xp GPU. With a lower resolution version, we achieve an mAP of 79.7 on VOC2007 with an inference speed of 13.0 milliseconds per image.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

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

In Computer Vision and Pattern Recognition

Citations in the corpus

Weight Standardization (2019, background): “Deep neural networks achieve state-of-the-arts in many vision tasks [4, 14, 20, 25, 31, 32, 39, 46].”
Mitigating adversarial effects through randomization (2017, background): “…2015; He et al., 2016a; Huang et al., 2016), object detection (Girshick et al., 2014; Girshick, 2015; Ren et al., 2015; Dai et al., 2016), semantic segmentation (Long et al., 2015; Chen et al., 2017; Zheng et al., 2015), visual concept discovery (Wang et al., 2015; Zhang et al., 2017) etc.”
IRS: A Large Synthetic Indoor Robotics Stereo Dataset for Disparity and Surface Normal Estimation (2019, background): “While monocular vision mainly focuses on core scene understanding tasks such as object detection [19], [20] and semantics segmentation [21], [22], stereo vision is popular to infer the spatial geometric information, including surface normal and depth/disparity, which play a significant role in the f”
Deep High-Resolution Representation Learning for Visual Recognition (2019, neutral): “DES [179] VGG16 512 32.”
AutoSlim: Towards One-Shot Architecture Search for Channel Numbers (2019, methodology): “This heuristic has been more-or-less used in followup network architecture designs including ResNets [13, 14], In-ception nets [15, 16, 17], MobileNets [6, 7] and networks for many vision applications [18, 19, 20, 21, 22].”
Object Detection in 20 Years: A Survey (2019, background): “To this end, some researchers have followed the idea of “weakly supervised learning”: instead of training based on pixel-wise annotation masks, they simply train the segmentation brunch based on the bounding-box level annotations [250, 271].”
Deep Regionlets for Object Detection (2017, methodology): “merits and properties. Extensive experiments on two detection benchmark datasets, PASCAL VOC [8] and Microsoft COCO [27] show that the proposed deep regionlet approach outperforms several competitors [36, 5, 47, 6, 32]. Even without segmentation labels and feature pyramid, we outperform state-of-the”
PCL: Proposal Cluster Learning for Weakly Supervised Object Detection (2018, background): “Recently, due to the development of Convolutional Neural Network (CNN) [1], [2] and the availability of large scale datasets with detailed boundingbox-level annotations [3], [4], [5], there have been great leap forwards in object detection [6], [7], [8], [9], [10], [11].”
Network Slimming by Slimmable Networks: Towards One-Shot Architecture Search for Channel Numbers (2019, methodology): “This heuristic has been more-or-less used in followup network architecture designs including ResNets [13, 14], In-ception nets [15, 16, 17], MobileNets [6, 7] and networks for many vision applications [18, 19, 20, 21, 22].”
IoU-balanced Loss Functions for Single-stage Object Detection (2019, background): “All these models can be classified into single-stage detectors [1, 2, 3, 4, 5, 6] and multi-stage detectors [7, 8, 9, 10, 11, 12, 13].”
Improving Transferability of Adversarial Examples With Input Diversity (2018, background): “Recent success of Convolutional Neural Networks (CNNs) leads to a dramatic performance improvement on various vision tasks, including image classification [14, 31, 12], object detection [9, 27, 39] and semantic segmentation [21, 4].”
QueryDet: Cascaded Sparse Query for Accelerating High-Resolution Small Object Detection (2021, unknown): “3 RefineDet512 [70] CVPR 2018 ResNet-101 36.”
Enriched Feature Guided Refinement Network for Object Detection (2019, methodology): “0 DES300 [42] VGG16 300× 300 79.”
Triply Supervised Decoder Networks for Joint Detection and Segmentation (2018, methodology): “(b) The branch for detection is refined by the branch for segmentation [35, 53].”
M3D-RPN: Monocular 3D Region Proposal Network for Object Detection (2019, background): “Despite that the RPN was conceived to be a preliminary stage within Faster R-CNN, it is often demonstrated to have promising effectiveness being extended to a single-shot standalone detector [25, 32, 38, 44].”
Efficient Featurized Image Pyramid Network for Single Shot Detector (2019, background): “[34] extended the standard SSD by integrating a semantic segmentation branch.”
ScratchDet: Exploring to Train Single-Shot Object Detectors from Scratch (2018, background): “Recent researches on one-stage approach focus on enriching features for detection (Fu et al. 2017), designing different architecture (Zhang et al. 2018) and addressing class imbalance issue (Zhang et al. 2017; Lin et al. 2017b).”
IoU-aware Single-stage Object Detector for Accurate Localization (2019, background): “Most of these models can be classified into single-stage object detectors [1, 2, 3, 4, 5, 6] and multi-stage object detectors [7, 8, 9, 10, 11, 12, 13].”
DetectoRS: Detecting Objects with Recursive Feature Pyramid and Switchable Atrous Convolution (2020, methodology): “There are two main categories of object detection methods: one-stage methods, e.g ., [45,50,56,60, 80, 81], and multi-stage methods, e.g ., [5, 7, 9, 25, 27, 58].”