Deep learning based real-time Industrial framework for rotten and fresh fruit detection using semantic segmentation
- PDF / 2,422,460 Bytes
- 11 Pages / 595.276 x 790.866 pts Page_size
- 121 Downloads / 257 Views
(0123456789().,-volV)(0123456789().,-volV)
TECHNICAL PAPER
Deep learning based real-time Industrial framework for rotten and fresh fruit detection using semantic segmentation Kyamelia Roy1
•
Sheli Sinha Chaudhuri1 • Sayan Pramanik2
Received: 6 November 2020 / Accepted: 16 November 2020 Ó Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Computer vision finds wide range of applications in fruit processing industries, allowing the tasks to be done with automation. Classification of fruit’s quality and thereby gradation of the same is very important for the industry manufacture unit for production of best quality finished food products and the finest quality of the raw fruits to be sellable in the market. In the present paper, detection of rotten or fresh apple has been accomplished based on the defects present on the peel of the fruit. The work proposes a semantic segmentation of the rotten portion present in the apple’s RGB image based on deep learning architecture. UNet and a modified version of it, the Enhanced UNet (En-UNet) are implemented for segmentation yielding promising results. The proposed En-UNet model generated enhanced outputs than UNet with training and validation accuracies of 97.46% and 97.54% respectively while UNet as the base architecture attaining an accuracy of 95.36%. The best mean IoU score under a threshold of 0.95 attained by En-UNet is 0.866 while that of UNet is 0.66. The experimental results show that the proposed model is a better one to be used for segmentation, detection and categorization of the rotten or fresh apples in real time.
1 Introduction Artificial intelligence (AI) with the aid of computer vision is boosting various sectors for quality production with high efficiency. In fields of agriculture and food industry, AI extends its help to the farmers and the manufacturers to improve their effectiveness and to overcome the traditional challenges under environmental hostile impacts. The adoption of AI in the agro-based industries has strengthened the technology to a greater extent. With the implementation of automation techniques, the food processing units have shown promising outcomes owing to excellent production and smart packaging. & Kyamelia Roy [email protected] Sheli Sinha Chaudhuri [email protected] Sayan Pramanik [email protected] 1
Department of Electronics and Telecommunication Engineering, Jadavpur University, Kolkata 700032, India
2
Department of Electronics and Communication Engineering, University of Engineering and Management, New Town, Kolkata 700160, India
In the twenty-first century, both fruit and food processing industries are undergoing soaring competitive positions. Global trade and market flow of fruits and vegetables determines geographical closeness between exporter and importers. In case of exporting or importing, there is a long and time-consuming process of transportation which causes hindrance in checking the quality of rotten or nearly rotten fruits amongst a bulk quantity of fruits. Thus produ
Data Loading...
