基于双分支网络的户外垃圾检测识别

doi:10.16180/j.cnki.issn1007-7820.2025.04.001

Abstract

Abstract:

The existing outdoor waste detection algorithms do not fully consider the advantages and disadvantages of CNN (Convolutional Neural Network) and Transformer in feature extraction, which limits the overall performance of the network. This study proposes a two-branch fusion network detection algorithm composed of CNN and Transformer. In the coding stage, a two-branch backbone network is constructed based on the advantages of CNN and Transformer to extract the feature information of the original image. Multi-scale convolutional module and multi-scale pooling module are used to eliminate the differences in dimension and semantics of extracted feature information, and the loss of detail information in deep neural network is reduced by strengthening feature extraction network. Six types of outdoor garbage images are collected, and a data set of garbage images with complex background is built to verify the performance of the proposed algorithm in outdoor garbage detection and recognition task. The experimental results show that the mAP(mean Average Precision) of the proposed algorithm on this data set is improved by about 5% when compared with the latest target detection algorithm. In order to verify the generalization performance of the proposed algorithm, a generalization experiment is carried out on the Huawei garbage classification challenge cup data set, and the experimental results show that mAP of the proposed algorithm is improved by about 2% when compared with the latest object detection algorithm.

Key words: garbage pollution, automatic detection and recognition, CNN, Transformer, dual branch fusion network, multi-scale convolution, multi-scale pooling, feature information

CLC Number:

TP391.4

ZHAO Wenqi, ZHANG Lixin, KAN Xi, ZHENG Haoren. Outdoor Garbage Detection and Recognition Based on Dual Branch Networks[J].Electronic Science and Technology, 2025, 38(4): 1-9.

Figures/Tables 16

Figure 1.

Table 1.

Structural parameters of dual branch fusion network"

	卷积分支	Transformer分支	解码器
L₁	1×1,32,1 3×3,64,2 1×1,32,1 1×1,16,1 3×3,32,1 1×1,32,1	$C o n v e m b e d (7 × 7,64, s t r i d e 2, p a d d i n g 3) T r a n s f o r m e r (d i m 64, h e a d)$	-
L₂	3×3,128,2 1×1,64,1 1×1,32,1 3×3,64,1 1×1,32,1 3×3,64,1 1×1,64,1	$C o n v e m b e d (7 × 7,128, s t r i d e 2, p a d d i n g 3) T r a n s f o r m e r (d i m 128, h e a d)$	3×3,128,1 1×1,9,1 3×3,128,1 1×1,1,1 1×1,4,1
L₃	3×3,256,2 1×1,128,1 1×1,64,1 3×3,128,1 1×1,64,1 3×3,128,1 1×1,128,1	$C o n v e m b e d (7 × 7,256, s t r i d e 2, p a d d i n g 3) T r a n s f o r m e r (d i m 256, h e a d)$	3×3,256,1 1×1,9,1 3×3,256,1 1×1,1,1 1×1,4,1
L₄	3×3,512,2 1×1,256,1 1×1,1 024,1 1×1,512,1 1×1,256,1 3×3,512,1 1×1,512,1	$C o n v e m b e d (7 × 7,512, s t r i d e 2, p a d d i n g 3) T r a n s f o r m e r (d i m 512, h e a d)$	3×3,512,1 1×1,9,1 3×3,512,1 1×1,1,1 1×1,4,1

Table 1.

Figure 2.

Figure 3.

Figure 4.

Figure 5.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Table 2.

Table 3.

Table 4.

Figure 11.

Table 5.

References 18

[1]	张伟, 刘娜, 江洋, 等. 基于YOLO神经网络的垃圾检测与分类[J]. 电子科技, 2022, 35(10):45-50.
	Zhang Wei, Liu Na, Jiang Yang, et al. Garbage detection and classification based on YOLO neural network[J]. Electronic Science and Technology, 2022, 35(10):45-50.
[2]	巨志勇, 张文馨, 翟春宇. 基于改进Canny算子的垃圾图像边缘检测[J]. 电子科技, 2020, 33(8):16-20.
	Ju Zhiyong, Zhang Wenxin, Zhai Chunyu. Garbage image edge detection based on improved Canny algorithm[J]. Electronic Science and Technology, 2020, 33(8):16-20.
[3]	Tao W Y, Jiang T P, Zheng Y R. Garbage classification and recognition system based on Faster R-CNN algorithm[J]. Network Security Technology and Application, 2020, 47(3):45-47.
[4]	Yang X, Zhao J, Zhao L, et al. Detection of river floating garbage based on improved YOLOv5[J]. Mathematics, 2022, 10(22):4366-4367.
[5]	Zhang P C, Zhao Q, Gao Z Y. Urban street garbage detection and cleanliness assessment approach fusing mobile edge computing and deep learning[J]. Journal of Chinese Computer Systems, 2019, 40(4):901-907.
[6]	Ma W, Wang X, Yu J. A lightweight feature fusion single shot multi box detector for garbage detection[J]. IEEE Access, 2020, 37(8):188577-188586.
[7]	Nagrath P, Jain R, Madan A, et al. SSDMNV2:A real time DNN-based face mask detection system using single shot multi box detector and MobileNetV2[J]. Sustainable Cities and Society, 2021, 66(1):102692-102693.
[8]	Chen W, Zhao Y, You T, et al. Automatic detection of scattered garbage regions using small unmanned aerial vehicle low-altitude remote sensing images for high-altitude natural reserve environmental protection[J]. Environmental Science and Technology, 2021, 55(6):3604-3611. doi: 10.1021/acs.est.0c04068 pmid: 33645984
[9]	Bai Y, Liu B, Li Y T. Research on recyclable garbage detection based on SSD algorithm[J]. Journal of Shaanxi University of Science and Technology, 2020, 38(6):152-158.
[10]	Xu W, Xiong W H, Yao J. Application of garbage detection based on improved YOLOv3 algorithm[J]. Journal of Optoelectronics·Laser, 2020, 31(9):928-938.
[11]	Yu L, Pan G, Li M. Garbage detection algorithm based on deep learning[C]. Chengdu:IEEE the Fifth Information Technology, Networking,Electronic and Automation Control Conference, 2021:469-473.
[12]	Gai R, Chen N, Yuan H. A detection algorithm for cherry fruits based on the improved YOLO-v4 model[J]. Neural Computing and Applications, 2023, 35(19):13895-13906.
[13]	Yan X, Yang Y, Feng L, et al. A garbage classification method based on improved YOLOv5[C]. Chengdu:International Conference on Networks,Communications and Information Technology,IEEE, 2022:1-5.
[14]	王莉, 何牧天, 徐硕, 等. 基于YOLOv5s网络的垃圾分类和检测[J]. 包装工程, 2021, 42(8):50-56.
	Wang Li, He Mutian, Xu Shuo, et al. Garbage classification and detection based on YOLOv5s network[J]. Packaging Engineering, 2021, 42(8):50-56.
[15]	Gm S P, Suraj K C. A novel CNN-based approach for intelligent garbage detection and localization[C]. Munich:The Fourth International Conference on Advances in Computing, Communication Control and Networking, 2022:784-789.
[16]	文灿华, 李佳, 董雪. 基于Faster RCNN的生活垃圾智能识别[J]. 激光与光电子学进展, 2020, 57(20):139-145.
	Wen Canhua, Li Jia, Dong Xue. Intelligent domestic garbage recognition based on Faster RCNN[J]. Laser and Optoelectronics Progress, 2020, 57(20):139-145.
[17]	Woo S, Park J, Lee J Y, et al. Cbam:Convolutional block attention module[C]. Munich: Proceedings of the European Conference on Computer Vision,2018:3-19.
[18]	Gupta T, Joshi R, Mukhopadhyay D, et al. A deep learning approach based hardware solution to categorise garbage in environment[J]. Complex and Intelligent Systems, 2022, 48(2):1-24.

算法模型	主干特征提取网络		特征交互模块		mAP/%
算法模型	CNN	Transformer	多尺度卷积模块	多尺度池化模块	mAP/%
CNN	√	×	×	×	87.30
Transformer	×	√	×	×	86.44
Combine	√	√	×	×	88.72
Combine-C	√	√	√	×	90.10
Combine-P	√	√	×	√	90.98
双分支	√	√	√	√	92.10

算法模型	mAP /%	AP/%
算法模型	mAP /%	塑料瓶	金属罐	纸箱	口罩	塑料袋	纸类
CNN	87.01	85.11	82.63	87.23	89.12	88.37	89.62
Transformer	87.15	84.60	81.79	88.91	89.89	89.23	88.48
Combine	89.37	87.20	84.89	90.04	91.92	91.16	91.06
Combine-C	90.55	88.77	86.65	91.20	92.63	91.83	92.26
Combine-P	90.12	88.26	85.73	90.78	92.15	91.25	92.59
双分支	92.10	90.22	87.23	92.56	94.04	92.18	93.71

算法模型	IoU0.5	IoU0.6	IoU0.7	时间/ms	FLOPs/10⁶	参数量/10⁶
算法模型	mAP/%	mAP/%	mAP/%	时间/ms	FLOPs/10⁶	参数量/10⁶
双分支	92.1	91.57	90.43	17.01	27 365.112	19.93
YOLOX	87.57	86.47	85.92	14.69	15 377.12	10.59
YOLOv5	76.18	75.59	74.77	16.87	8 222.87	7.09
YOLOv4	73.65	72.83	71.92	17.66	70 788.35	63.98
YOLOv3	64.49	63.92	62.82	19.48	77 607.16	61.57
Faster R-CNN	69.63	68.88	67.91	148.94	473 284.69	28.36
Efficientdet	65.77	64.71	63.98	17.71	3 624.17	3.83
Retinanet	62.88	62.17	61.23	116.72	77 966.19	30.04
Centernet	66.51	65.72	64.82	20.23	54 650.8	32.66

算法模型	mAP/%	RC/%	PR/%	AP/%
算法模型	mAP/%	RC/%	PR/%	毛绒玩具	旧衣服	茶叶	包	烟头
Faster R-CNN	58.67	53.76	57.92	61.82	56.92	64.68	57.21	51.97
YOLOv3	60.83	56.27	60.54	64.60	58.71	67.52	59.34	54.73
YOLOv4	62.12	57.83	61.71	64.57	60.34	68.39	60.84	55.34
YOLOv5	62.77	58.29	62.22	65.34	60.85	69.20	62.17	55.72
YOLOX	63.52	58.94	63.81	66.05	61.57	69.37	62.43	56.18
双分支	65.71	60.83	65.39	68.27	62.41	70.73	64.12	59.78

Outdoor Garbage Detection and Recognition Based on Dual Branch Networks

RichHTML

PDF (PC)

Like

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 16

References 18

Related Articles 15

Metrics

Comments

Recommended 0

[1]	BI Hanjia, YANG Churui, WANG Xiaoyu, HUANG Yuehua. Multi-Class Defect Target Detection for Transmission Lines Based on Improved YOLOv7 [J]. Electronic Science and Technology, 2025, 38(4): 16-24.
[2]	LI Linjie, LIU Deer. Research on Urban Waterlogging Simulation and Monitoring Platform Based on Container Service [J]. Electronic Science and Technology, 2025, 38(4): 52-58.
[3]	XU Yaobo, YANG Xinqiang, XU Guangchao, YANG Shihao, DUAN Guoyong. Transformer Fault Identification Method Based on Gramian Angle Difference Field and CNN-BiGRU [J]. Electronic Science and Technology, 2025, 38(4): 73-79.
[4]	CHEN Yuyang, LI Feng. Integration of CNN and Transformer for Retinal OCT Image Fluid Segmentation Method [J]. Electronic Science and Technology, 2025, 38(3): 47-59.
[5]	ZHOU Bin, KAN Jiarong, CHEN Heming, CHEN Weiwei, LI Yan, XU Sudong. A Transformerless Half-Bridge Lithium Battery Equalizer Based on Phase-Shift Strategy [J]. Electronic Science and Technology, 2025, 38(2): 1-9.
[6]	ZHENG Fangliang, WANG Yannian, LIAN Jihong, RUAN Pei. Face Image Super-Resolution Reconstruction Based on Conditional Priori Swin Transformer [J]. Electronic Science and Technology, 2025, 38(2): 35-41.
[7]	XIONG Zhangliang, CHEN Suting, XUAN Zhibin, ZHAO Tingchen, LIU Jiajun, FU Mannan. Design of High Speed Encoding and Decoding Circuit for Isolation Driver Chip [J]. Electronic Science and Technology, 2025, 38(2): 84-92.
[8]	HE Zhiqiang, SUN Zhanquan. Swin-Transformer-Based Carotid Ultrasound Image Plaque Segmentation [J]. Electronic Science and Technology, 2024, 37(9): 48-56.
[9]	ZHU Yong, HUANG Yongming, HE Xing. Research on Object Detection Based on Radar and Video Fusion [J]. Electronic Science and Technology, 2024, 37(8): 1-7.
[10]	TONG Zhaojing, JING Lifei, LAN Mengyue. A Bayesian Network Optimization Method for Transformer Fault Diagnosis [J]. Electronic Science and Technology, 2024, 37(8): 34-39.
[11]	LIANG Qiyu, WANG Yonggang, QIU Shengshun. Overvoltage Suppression of Single Switch Resonant Pulse Power Supply [J]. Electronic Science and Technology, 2024, 37(6): 17-28.
[12]	XIA Rongcheng, LIU Deer. Vehicle Detection and Analysis in Urban Waterlogging Area Based on Deep Learning [J]. Electronic Science and Technology, 2024, 37(5): 18-24.
[13]	TONG Zhaojing, LAN Mengyue, JING Lifei. Research on Transformer Fault Diagnosis Based on Improved Bayesian Network [J]. Electronic Science and Technology, 2024, 37(5): 47-53.
[14]	PANG Jiangfei, SUN Zhanquan. Multi-Encoder Transformer for End-to-End Speech Recognition [J]. Electronic Science and Technology, 2024, 37(4): 1-7.
[15]	LIANG Chenye, ZHANG Xuanxiong. Research on Multiclass Garbage Classification Algorithm Based on Improved MobileNet Network [J]. Electronic Science and Technology, 2024, 37(4): 38-46.