See the published version of this article at EURASIP Journal on Image and Video Processing.
This is a preprint, a preliminary version of a manuscript that has not completed peer review at a journal. Research Square does not conduct peer review prior to posting preprints. The posting of a preprint on this server should not be interpreted as an endorsement of its validity or suitability for dissemination as established information or for guiding clinical practice.
Learn more about In Review
Research
Anchor free object detection with mask attention
Beibei Fan
Shanghai University
Corresponding Author
License:
This work is licensed under a CC BY 4.0 License. Read Full License
View the published version at EURASIP Journal on Image and Video Processing.
The anchor-free method based on key point detection has made great progress. However, the anchor-free method is too dependent on using a convolutional network to generate a rough heat map. This is difficult to detect for objects with a large size variation and dense and overlapping objects. To solve this problem, first, we propose a mask attention mechanism for object detection methods. And make full use of the advantages of the attention mechanism to improve the accuracy of network detection heat map generation. Then, we designed an optimized fire model to reduce the size of the model. The fire model is an extension of grouped convolution. The fire model allows each group of convolutional network features to learn the same feature through purposeful grouping. In this paper, the mask attention mechanism uses object segmentation images to guide the generation of corner heat maps. Our approach achieved an accuracy of 91.84% and a recall 89.83% in the Tencent-100K dataset. Compared with the popular object detection methods, the proposed method has advantages in model size and accuracy.
Keywords
object detection, anchor-free, convolutional neural network, attention mechanism
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Due to technical limitations, full-text HTML conversion of this manuscript could not be completed. However, the manuscript can be downloaded and accessed as a PDF.
Badges
Prescreen
This manuscript passed our Prescreen assessment
Complete author information
Appropriate declaration statements
Potential risks to human health
Prescreen
Peer Review Timeline
CURRENT STATUS: Published
Version 2
Posted 03 Jun, 2020
Published
On 14 Jul, 2020
No community comments so far
Editorial decision: Accept
On 03 Jul, 2020
Review #1 received
Received 25 Jun, 2020
Review #2 received
Received 25 Jun, 2020
Review #3 received
Received 18 Jun, 2020
Review #4 received
Received 18 Jun, 2020
Reviewer #4 agreed
On 04 Jun, 2020
Reviewers invited
Invitations sent on 31 May, 2020
Reviewer #1 agreed
On 31 May, 2020
Reviewer #2 agreed
On 31 May, 2020
Reviewer #3 agreed
On 31 May, 2020
Editor assigned
On 28 May, 2020
Submission checks complete
On 27 May, 2020
Editor invited
On 27 May, 2020
No comments provided
Review #4 received
Received 26 Apr, 2020
Editorial decision: Major revision
On 26 Apr, 2020
Review #3 received
Received 25 Apr, 2020
Reviewer #10 agreed
On 04 Apr, 2020
Reviewer #9 agreed
On 01 Apr, 2020
Reviewer #6 agreed
On 31 Mar, 2020
Reviewer #7 agreed
On 31 Mar, 2020
Reviewer #8 agreed
On 31 Mar, 2020
Review #2 received
Received 30 Mar, 2020
Reviewer #5 agreed
On 28 Mar, 2020
Reviewer #4 agreed
On 28 Mar, 2020
Reviewer #2 agreed
On 27 Mar, 2020
Reviewer #3 agreed
On 27 Mar, 2020
Review #1 received
Received 27 Mar, 2020
Reviewers invited
Invitations sent on 24 Mar, 2020
Reviewer #1 agreed
On 24 Mar, 2020
Editor assigned
On 27 Feb, 2020
Submission checks complete
On 26 Feb, 2020
Editor invited
On 26 Feb, 2020
First submitted
On 24 Feb, 2020
Metrics
Comments: 0
PDF Downloads: ...
HTML Views: ...
Subject Areas
Mechanical Engineering
Learn more about our company.
This preprint is under consideration at EURASIP Journal on Image and Video Processing. A preprint is a preliminary version of a manuscript that has not completed peer review at a journal. Research Square does not conduct peer review prior to posting preprints. The posting of a preprint on this server should not be interpreted as an endorsement of its validity or suitability for dissemination as established information or for guiding clinical practice.
Learn more about In Review
Research
Anchor free object detection with mask attention
Beibei Fan
Shanghai University
Corresponding Author
Badges
Prescreen
This manuscript passed our Prescreen assessment
Complete author information
Appropriate declaration statements
Potential risks to human health
Prescreen
Peer Review Timeline
CURRENT STATUS: Published
Version 2
Posted 03 Jun, 2020
Published
On 14 Jul, 2020
No community comments so far
Editorial decision: Accept
On 03 Jul, 2020
Review #1 received
Received 25 Jun, 2020
Review #2 received
Received 25 Jun, 2020
Review #3 received
Received 18 Jun, 2020
Review #4 received
Received 18 Jun, 2020
Reviewer #4 agreed
On 04 Jun, 2020
Reviewers invited
Invitations sent on 31 May, 2020
Reviewer #1 agreed
On 31 May, 2020
Reviewer #2 agreed
On 31 May, 2020
Reviewer #3 agreed
On 31 May, 2020
Editor assigned
On 28 May, 2020
Submission checks complete
On 27 May, 2020
Editor invited
On 27 May, 2020
No comments provided
Review #4 received
Received 26 Apr, 2020
Editorial decision: Major revision
On 26 Apr, 2020
Review #3 received
Received 25 Apr, 2020
Reviewer #10 agreed
On 04 Apr, 2020
Reviewer #9 agreed
On 01 Apr, 2020
Reviewer #6 agreed
On 31 Mar, 2020
Reviewer #7 agreed
On 31 Mar, 2020
Reviewer #8 agreed
On 31 Mar, 2020
Review #2 received
Received 30 Mar, 2020
Reviewer #5 agreed
On 28 Mar, 2020
Reviewer #4 agreed
On 28 Mar, 2020
Reviewer #2 agreed
On 27 Mar, 2020
Reviewer #3 agreed
On 27 Mar, 2020
Review #1 received
Received 27 Mar, 2020
Reviewers invited
Invitations sent on 24 Mar, 2020
Reviewer #1 agreed
On 24 Mar, 2020
Editor assigned
On 27 Feb, 2020
Submission checks complete
On 26 Feb, 2020
Editor invited
On 26 Feb, 2020
First submitted
On 24 Feb, 2020
Metrics
Comments: 0
PDF Downloads: ...
HTML Views: ...
Subject Areas
Mechanical Engineering
License:
This work is licensed under a CC BY 4.0 License. Read Full License
View the published version at EURASIP Journal on Image and Video Processing.
The anchor-free method based on key point detection has made great progress. However, the anchor-free method is too dependent on using a convolutional network to generate a rough heat map. This is difficult to detect for objects with a large size variation and dense and overlapping objects. To solve this problem, first, we propose a mask attention mechanism for object detection methods. And make full use of the advantages of the attention mechanism to improve the accuracy of network detection heat map generation. Then, we designed an optimized fire model to reduce the size of the model. The fire model is an extension of grouped convolution. The fire model allows each group of convolutional network features to learn the same feature through purposeful grouping. In this paper, the mask attention mechanism uses object segmentation images to guide the generation of corner heat maps. Our approach achieved an accuracy of 91.84% and a recall 89.83% in the Tencent-100K dataset. Compared with the popular object detection methods, the proposed method has advantages in model size and accuracy.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Due to technical limitations, full-text HTML conversion of this manuscript could not be completed. However, the manuscript can be downloaded and accessed as a PDF.