This content is not included in
your SAE MOBILUS subscription, or you are not logged in.
The Effect of Unfine-Tuned Super-Resolution Networks Act on Object Detection
Technical Paper
2020-01-5034
ISSN: 0148-7191, e-ISSN: 2688-3627
This content contains downloadable datasets
Annotation ability available
Sector:
Language:
English
Abstract
In order to explore approaches for improving object detection accuracy in intelligent vehicle system, we exploit super-resolution techniques. A novel method is proposed to confirm the conjecture whether some popular super-resolution networks used for environmental perception of intelligent vehicles and robots can indeed improve the detection accuracy. COCO dataset which contains images from complex ordinary environment is utilized for the verification experiment, due to it can adequately verify the generalization of each algorithm and the consistency of experimental results. Using two representative object detection networks to produce the detection results, namely Faster R-CNN and YOLOv3, we devise to reduce the impact of resizing operation. The two networks allow us to compare the performance of object detection between using original and super-resolved images. We quantify the effect of each super-resolution techniques as well. Shown from our experimental results, the super-resolution networks provide average 3%-12% decrease in mAP (mean Average Precision) on enhancement level of ×2 and average 8%-19% decrease on enlargement level of ×4 compared to the conventional methods.
Recommended Content
Technical Paper | Moving Obstacle Detection from Moving Platforms |
Technical Paper | Desert Research and Technology Studies 2006 Report |
Technical Paper | Evaluation of Advanced Machine-Vision Sensors for Measuring the Tread Depth of a Tire |
Authors
Citation
Zhenwen, D., Lu, X., Guang, C., Zehan, W. et al., "The Effect of Unfine-Tuned Super-Resolution Networks Act on Object Detection," SAE Technical Paper 2020-01-5034, 2020, https://doi.org/10.4271/2020-01-5034.Data Sets - Support Documents
Title | Description | Download |
---|---|---|
Unnamed Dataset 1 | ||
Unnamed Dataset 2 | ||
Unnamed Dataset 3 | ||
Unnamed Dataset 4 | ||
Unnamed Dataset 5 | ||
Unnamed Dataset 6 |
Also In
References
- Kaiming , H. , Xiangyu , Z. , Shaoqing , R. et al. 2015
- Alice , J. , Jonathon , P. , Fang , J. et al. Face Recognition Algorithms Surpass Humans Matching Faces over Changes in Illumination IEEE Trans. Pattern Analysis and Machine Intelligence 29 9 1642 1646 2007
- Feng , D. , Hasse , C. , Rosenbaum , L. et al. 2019
- Ren , S. , He , K. , Girshick , R. et al. Faster r-Cnn: Towards Real-Time Object Detection with Region Proposal Networks NIPS 2015
- Redmon , J. , Farhadi A. 2018
- Wenming , Y. , Xuechen , Z. , Yapeng , T. et al. Deep Learning for Single Image Super-Resolution: A Brief Review 2018
- Chao , D. , Chen , C. , Kaiming , H. et al. 2014
- Guido , M. , Razvan , P. , and Yoshua , B. On the Number of Linear Regions of Deep Neural Networks NIPS 2014
- Kim , J. , Kwon , J. , and Mu , K. Accurate Image Super-Resolution Using Very Deep Convolutional Networks Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 2016 1646 1654
- Kaiming , H. , Xiangyu , Z. , Shaoqing , R. , and Sun , J. 2015
- Lim , B. , Son , S. , Kim , H. et al. 2017
- Xintao , W. , Ke , Y. , Shixiang , W. et al. ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks 2018
- Zhongqiu , Z. , Peng , Z. , Shou-tao , X. et al. Object Detection with Deep Learning: A Review 2018
- Lin , T. , Maire , M. , Belongie , S. et al. Microsoft Coco: Common Objects in Context 740 755
- Muhammad , H. , Greg , S. , and Norimichi , U. 2018