Enabling Deep Residual Networks for Weakly Supervised Object Detection

Shen, Yunhang; Ji, Rongrong; Wang, Yan; Chen, Zhiwei; Zheng, Feng; Huang, Feiyue; Wu, Yunsheng

doi:10.1007/978-3-030-58598-3_8

Cited by 34 publications

(14 citation statements)

References 69 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We use ResNet50 as the backbone by default and also report the results using VGG16. We observe that the results using VGG16 are only slightly worse than those using ResNet50, which is consistent with the observation in [33]. One possible explanation is that the MIL classifier may back-propagate uncertain and inaccurate gradients to backbones while skip connection in ResNet50 can not alleviate this issue.…”

Section: Experiments On Coco-60supporting

confidence: 87%

Mixed Supervised Object Detection by Transferring Mask Prior and Semantic Similarity

Liu¹,

Zhang²,

Niu³

et al. 2021

Preprint

View full text Add to dashboard Cite

Object detection has achieved promising success, but requires large-scale fullyannotated data, which is time-consuming and labor-extensive. Therefore, we consider object detection with mixed supervision, which learns novel object categories using weak annotations with the help of full annotations of existing base object categories. Previous works using mixed supervision mainly learn the classagnostic objectness from fully-annotated categories, which can be transferred to upgrade the weak annotations to pseudo full annotations for novel categories. In this paper, we further transfer mask prior and semantic similarity to bridge the gap between novel categories and base categories. Specifically, the ability of using mask prior to help detect objects is learned from base categories and transferred to novel categories. Moreover, the semantic similarity between objects learned from base categories is transferred to denoise the pseudo full annotations for novel categories. Experimental results on three benchmark datasets demonstrate the effectiveness of our method over existing methods. Codes are available at https://github.com/bcmi/TraMaS-Weak-Shot-Object-Detection.

show abstract

Section: Experiments On Coco-60supporting

confidence: 87%

Mixed Supervised Object Detection by Transferring Mask Prior and Semantic Similarity

Liu¹,

Zhang²,

Niu³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Training the semi-autonomous learning network with ResNet backbone will reduce the identification of the proposed features, and will be weak in localizing object instances. Discovered by [36], the proposed semi-autonomous learning algorithm in this paper takes MRN as the backbone network.…”

Section: Modified Residual Network For Semi-autonomous Learningmentioning

confidence: 99%

“…This is because non-maximum down-sampling may not retain the activation and gradient of the information flowing through the network under weak supervision. Inspired by [36], a MRN is proposed in this paper, and the small kernel convolution and max-pooling are used to improve the robustness of information flow, which makes the object boundary more detailed. Specifically, the original stem block is replaced by three conservative 3 × 3 convolutions, and the first and third convolutions are followed by the max-pooling layer.…”

Section: mentioning

confidence: 99%

See 1 more Smart Citation

Semi-Autonomous Learning Algorithm for Remote Image Object Detection Based on Aggregation Area Instance Refinement

Cheng

et al. 2021

Remote Sensing

View full text Add to dashboard Cite

Semi-autonomous learning for object detection has attracted more and more attention in recent years, which usually tends to find only one object instance with the highest score in each image. However, this strategy usually highlights the most representative part of the object instead of the whole object, which may lead to the loss of a lot of important information. To solve this problem, a novel end-to-end aggregate-guided semi-autonomous learning residual network is proposed to perform object detection. Firstly, a progressive modified residual network (MRN) is applied to the backbone network to make the detector more sensitive to the boundary features of the object. Then, an aggregate-based region-merging strategy (ARMS) is designed to select high-quality instances by selecting aggregation areas and merging these regions. The ARMS selects the aggregation areas that are highly related to the object through association coefficient, and then evaluates the aggregation areas through a similarity coefficient and fuses them to obtain high-quality object instance areas. Finally, a regression-locating branch is further developed to refine the location of the object, which can be optimized jointly with regional classification. Extensive experiments demonstrate that the proposed method is superior to state-of-the-art methods.

show abstract

“…In this case, earlier detection algorithms will be more complicated in design, and the detection effect cannot meet the actual demand. After more than a decade of development, deep neural networks have gradually matured, and many high-level network design solutions have emerged, becoming the mainstream algorithm for solving object detection problems [10][11][12][13][14][15][16][17][18][19][20]. Among these methods, Faster-RCNN [10] provides a new idea to accomplish the task of multi-category target detection for images on an efficient and high accuracy basis.…”

Section: Introductionmentioning

confidence: 99%

Multi-Stage Feature Enhancement Pyramid Network for Detecting Objects in Optical Remote Sensing Images

Zhang

Shen

2022

Remote Sensing

View full text Add to dashboard Cite

The intelligent detection of objects in remote sensing images has gradually become a research hotspot for experts from various countries, among which optical remote sensing images are considered to be the most important because of the rich feature information, such as the shape, texture and color, that they contain. Optical remote sensing image target detection is an important method for accomplishing tasks, such as land use, urban planning, traffic guidance, military monitoring and maritime rescue. In this paper, a multi stages feature pyramid network, namely the Multi-stage Feature Enhancement Pyramid Network (Multi-stage FEPN), is proposed, which can effectively solve the problems of blurring of small-scale targets and large scale variations of targets detected in optical remote sensing images. The Content-Aware Feature Up-Sampling (CAFUS) and Feature Enhancement Module (FEM) used in the network can perfectly solve the problem of fusion of adjacent-stages feature maps. Compared with several representative frameworks, the Multi-stage FEPN performs better in a range of common detection metrics, such as model accuracy and detection accuracy. The mAP reaches 0.9124, and the top-1 detection accuracy reaches 0.921 on NWPU VHR-10. The results demonstrate that Multi-stage FEPN provides a new solution for the intelligent detection of targets in optical remote sensing images.

show abstract

Enabling Deep Residual Networks for Weakly Supervised Object Detection

Cited by 34 publications

References 69 publications

Mixed Supervised Object Detection by Transferring Mask Prior and Semantic Similarity

Mixed Supervised Object Detection by Transferring Mask Prior and Semantic Similarity

Semi-Autonomous Learning Algorithm for Remote Image Object Detection Based on Aggregation Area Instance Refinement

Multi-Stage Feature Enhancement Pyramid Network for Detecting Objects in Optical Remote Sensing Images

Contact Info

Product

Resources

About