Domestic garbage recognition and detection based on Faster R-CNN

Nie, Zhifeng; Duan, Wenjie; Li, Xiangdong

doi:10.1088/1742-6596/1738/1/012089

Cited by 11 publications

(7 citation statements)

References 1 publication

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The image processing techniques used in this work were chosen so they could be tunable and reliably programmed in hardware, without sacrificing the overall performance. Some works use deep learning approaches (DL) 16 , 19 , which must be trained with a great number of images. Although, multiple resources are available on the internet, a customized DL solution needs the labeling of hundreds or perhaps thousands of images, which involves time and resources that are not available at the time of the system development.…”

Section: Resultsmentioning

confidence: 99%

Implementation and integration of image processing blocks in a real-time bottle classification system

Jimenez-Arteaga

Barrón-Zambrano

Hernández-Mier³

et al. 2022

Sci Rep

View full text Add to dashboard Cite

A practical solution to the problems caused by the water, air, and soil pollution produced by the large volumes of waste is recycling. Plastic and glass bottle recycling is a practical solution but sometimes unfeasible in underdeveloped countries. In this paper, we propose a high-performance real-time hardware architecture for bottle classification, that process input image bottles to generate a bottle color as output. The proposed architecture was implemented on a Spartan-6 Field Programmable Gate Array, using a Hardware Description Language. The proposed system was tested for several input resolutions up to 1080 p, but it is flexible enough to support input video resolutions up to 8 K. There is no evidence of a high-performance bottle classification system in the state-of-the-art. The main contribution of this paper is the implementation and integration of a set of dedicated image processing blocks in a high-performance real-time bottle classification system. These hardware modules were integrated into a compact and tunable architecture, and was tested in a simulated environment. Concerning the image processing algorithm implemented in the FPGA, the maximum processing rate is 60 frames per second. In practice, the maximum number of bottles that can be processed would be limited by the mechanical aspects of the bottle transportation system.

show abstract

Section: Resultsmentioning

confidence: 99%

Implementation and integration of image processing blocks in a real-time bottle classification system

Jimenez-Arteaga

Barrón-Zambrano

Hernández-Mier³

et al. 2022

Sci Rep

View full text Add to dashboard Cite

show abstract

“…In Ref. [9], for instance, Nie et al applied Faster RCNN with the backbone of ResNet-50 to detect 3984 garbage images. The results show that the accuracy of garbage recognition is 89.68%, which is nearly 10 percentage points ahead of the compared traditional approaches.…”

Section: Feature Extraction and Object Recognition Algorithm Based On...mentioning

confidence: 99%

“…Existing approaches [6][7][8][9] mainly focus on how to quickly and accurately distinguish the categories and find the locations of target garbage in images. Many researchers have also proposed garbage classification tools, such as smart trash can [10][11][12][13] .…”

Section: Introductionmentioning

confidence: 99%

New Benchmark for Household Garbage Image Recognition

Huanyi

et al. 2022

Tsinghua Sci. Technol.

View full text Add to dashboard Cite

Household garbage images are usually faced with complex backgrounds, variable illuminations, diverse angles, and changeable shapes, which bring a great difficulty in garbage image classification. Due to the ability to discover problem-specific features, deep learning and especially convolutional neural networks (CNNs) have been successfully and widely used for image representation learning. However, available and stable household garbage datasets are insufficient, which seriously limits the development of research and application. Besides, the state-of-the-art in the field of garbage image classification is not entirely clear. To solve this problem, in this study, we built a new open benchmark dataset for household garbage image classification by simulating different lightings, backgrounds, angles, and shapes. This dataset is named 30 classes of household garbage images (HGI-30), which contains 18 000 images of 30 household garbage classes. The publicly available HGI-30 dataset allows researchers to develop accurate and robust methods for household garbage recognition. We also conducted experiments and performance analyses of the state-of-the-art deep CNN methods on HGI-30, which serves as baseline results on this benchmark.

show abstract

“…Mature CNNs, like faster region based CNN ( Faster RCNN) [39], Mask RCNN [40], and deformable convolutional network (DCN) [41], are outstanding representatives of the application of region proposals, which are often referred to as two-stage algorithms. In [42], for instance, Nie et al applied Faster RCNN with the backbone of ResNet-50 to detect 3,984 garbage images. The results show that the accuracy of garbage recognition is 89.68%, which is nearly 10 percentage points ahead of the compared traditional approaches.…”

Section: Feature Extraction and Object Recognition Algorithm Based On...mentioning

confidence: 99%

“…Existing approaches [6,7,8,9] mainly focus on how to quickly and accurately distinguish the categories and find the locations of target garbage in images. Many researchers have also proposed garbage classification tools, such as smart trash can [10,11,12,13].…”

Section: Introductionmentioning

confidence: 99%

New Benchmark for Household Garbage Image Recognition

Wu¹,

Huanyi²,

Wu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Household garbage images are usually faced with complex backgrounds, variable illuminations, diverse angles, and changeable shapes, which bring a great difficulty in garbage image classification. Due to the ability to discover problem-specific features, deep learning and especially convolutional neural networks (CNNs) have been successfully and widely used for image representation learning. However, available and stable household garbage datasets are insufficient, which seriously limits the development of research and application. Besides, the state of the art in the field of garbage image classification is not entirely clear. To solve this problem, in this study, we built a new open benchmark dataset for household garbage image classification by simulating different lightings, backgrounds, angles, and shapes. This dataset is named 30 Classes of Household Garbage Images (HGI-30), which contains 18,000 images of 30 household garbage classes. The publicly available HGI-30 dataset allows researchers to develop accurate and robust methods for household garbage recognition. We also conducted experiments and performance analysis of the state-of-the-art deep CNN methods on HGI-30, which serves as baseline results on this benchmark.

show abstract

Domestic garbage recognition and detection based on Faster R-CNN

Cited by 11 publications

References 1 publication

Implementation and integration of image processing blocks in a real-time bottle classification system

Implementation and integration of image processing blocks in a real-time bottle classification system

New Benchmark for Household Garbage Image Recognition

New Benchmark for Household Garbage Image Recognition

Contact Info

Product

Resources

About