Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection
The development of object detection technology makes it possible for robots to interact with people and the environment, but the changeable application scenarios make the detection accuracy of small and medium objects in the practical application of object detection technology low. In this paper, ba...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2022-05-01
|
| Series: | Frontiers in Neurorobotics |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fnbot.2022.881021/full |
| _version_ | 1817991140756946944 |
|---|---|
| author | Li Huang Li Huang Cheng Chen Juntong Yun Juntong Yun Ying Sun Ying Sun Ying Sun Jinrong Tian Jinrong Tian Zhiqiang Hao Zhiqiang Hao Zhiqiang Hao Hui Yu Hongjie Ma |
| author_facet | Li Huang Li Huang Cheng Chen Juntong Yun Juntong Yun Ying Sun Ying Sun Ying Sun Jinrong Tian Jinrong Tian Zhiqiang Hao Zhiqiang Hao Zhiqiang Hao Hui Yu Hongjie Ma |
| author_sort | Li Huang |
| collection | DOAJ |
| description | The development of object detection technology makes it possible for robots to interact with people and the environment, but the changeable application scenarios make the detection accuracy of small and medium objects in the practical application of object detection technology low. In this paper, based on multi-scale feature fusion of indoor small target detection method, using the device to collect different indoor images with angle, light, and shade conditions, and use the image enhancement technology to set up and amplify a date set, with indoor scenarios and the SSD algorithm in target detection layer and its adjacent features fusion. The Faster R-CNN, YOLOv5, SSD, and SSD target detection models based on multi-scale feature fusion were trained on an indoor scene data set based on transfer learning. The experimental results show that multi-scale feature fusion can improve the detection accuracy of all kinds of objects, especially for objects with a relatively small scale. In addition, although the detection speed of the improved SSD algorithm decreases, it is faster than the Faster R-CNN, which better achieves the balance between target detection accuracy and speed. |
| first_indexed | 2024-04-14T01:09:24Z |
| format | Article |
| id | doaj.art-7bad3a57af96486f9fc487b004ca10d5 |
| institution | Directory Open Access Journal |
| issn | 1662-5218 |
| language | English |
| last_indexed | 2024-04-14T01:09:24Z |
| publishDate | 2022-05-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Neurorobotics |
| spelling | doaj.art-7bad3a57af96486f9fc487b004ca10d52022-12-22T02:21:09ZengFrontiers Media S.A.Frontiers in Neurorobotics1662-52182022-05-011610.3389/fnbot.2022.881021881021Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target DetectionLi Huang0Li Huang1Cheng Chen2Juntong Yun3Juntong Yun4Ying Sun5Ying Sun6Ying Sun7Jinrong Tian8Jinrong Tian9Zhiqiang Hao10Zhiqiang Hao11Zhiqiang Hao12Hui Yu13Hongjie Ma14College of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan, ChinaHubei Province Key Laboratory of Intelligent Information Processing and Real-Time Industrial System, Wuhan University of Science and Technology, Wuhan, ChinaCollege of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan, ChinaKey Laboratory of Metallurgical Equipment and Control Technology of Ministry of Education, Wuhan University of Science and Technology, Wuhan, ChinaHubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering, Wuhan University of Science and Technology, Wuhan, ChinaKey Laboratory of Metallurgical Equipment and Control Technology of Ministry of Education, Wuhan University of Science and Technology, Wuhan, ChinaHubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering, Wuhan University of Science and Technology, Wuhan, ChinaPrecision Manufacturing Research Institute, Wuhan University of Science and Technology, Wuhan, ChinaKey Laboratory of Metallurgical Equipment and Control Technology of Ministry of Education, Wuhan University of Science and Technology, Wuhan, ChinaHubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering, Wuhan University of Science and Technology, Wuhan, ChinaKey Laboratory of Metallurgical Equipment and Control Technology of Ministry of Education, Wuhan University of Science and Technology, Wuhan, ChinaHubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering, Wuhan University of Science and Technology, Wuhan, ChinaPrecision Manufacturing Research Institute, Wuhan University of Science and Technology, Wuhan, ChinaSchool of Creative Technologies, University of Portsmouth, Portsmouth, United KingdomSchool of Energy and Electronic Engineering, University of Portsmouth, Portsmouth, United KingdomThe development of object detection technology makes it possible for robots to interact with people and the environment, but the changeable application scenarios make the detection accuracy of small and medium objects in the practical application of object detection technology low. In this paper, based on multi-scale feature fusion of indoor small target detection method, using the device to collect different indoor images with angle, light, and shade conditions, and use the image enhancement technology to set up and amplify a date set, with indoor scenarios and the SSD algorithm in target detection layer and its adjacent features fusion. The Faster R-CNN, YOLOv5, SSD, and SSD target detection models based on multi-scale feature fusion were trained on an indoor scene data set based on transfer learning. The experimental results show that multi-scale feature fusion can improve the detection accuracy of all kinds of objects, especially for objects with a relatively small scale. In addition, although the detection speed of the improved SSD algorithm decreases, it is faster than the Faster R-CNN, which better achieves the balance between target detection accuracy and speed.https://www.frontiersin.org/articles/10.3389/fnbot.2022.881021/fullindoor scenesmall target detectionconvolutional neural networkmulti-scale feature fusionSSD |
| spellingShingle | Li Huang Li Huang Cheng Chen Juntong Yun Juntong Yun Ying Sun Ying Sun Ying Sun Jinrong Tian Jinrong Tian Zhiqiang Hao Zhiqiang Hao Zhiqiang Hao Hui Yu Hongjie Ma Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection Frontiers in Neurorobotics indoor scene small target detection convolutional neural network multi-scale feature fusion SSD |
| title | Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection |
| title_full | Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection |
| title_fullStr | Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection |
| title_full_unstemmed | Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection |
| title_short | Multi-Scale Feature Fusion Convolutional Neural Network for Indoor Small Target Detection |
| title_sort | multi scale feature fusion convolutional neural network for indoor small target detection |
| topic | indoor scene small target detection convolutional neural network multi-scale feature fusion SSD |
| url | https://www.frontiersin.org/articles/10.3389/fnbot.2022.881021/full |
| work_keys_str_mv | AT lihuang multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT lihuang multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT chengchen multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT juntongyun multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT juntongyun multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT yingsun multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT yingsun multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT yingsun multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT jinrongtian multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT jinrongtian multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT zhiqianghao multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT zhiqianghao multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT zhiqianghao multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT huiyu multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection AT hongjiema multiscalefeaturefusionconvolutionalneuralnetworkforindoorsmalltargetdetection |