Vision-Based Object Classification Using Deep Learning for Mixed Palletizing Operation in an Automated Warehouse Environment

Patel, AD and Chowdhury, AR (2022) Vision-Based Object Classification Using Deep Learning for Mixed Palletizing Operation in an Automated Warehouse Environment. In: 2nd International Conference on Recent Advances in Manufacturing, RAM 2021, 10 - 12 June 2021, Surat, pp. 991-1011.

PDF RAM_2021.pdf - Published Version Restricted to Registered users only Download (733kB) | Request a copy

Abstract

Present object identification and classification methods in warehouse automation either require a tedious calibration process or run at a speed that is slow for real time application. This research paper proposes a computer vision-based method using a deep convolutional neural learning network to calculate dimensions of 5 distinct objects found in an automated robot warehouse and classify them using images taken from a camera for the purpose of mixed palletizing inside warehouses. It achieved a classification accuracy of 98.94 on the machine parts images dataset created by the authors comprising more than 1450 images by utilizing transfer learning and data augmentation. The employed method ResNet-50 uses a combination of residual learning and deep architecture parsing called residual network architecture. The proposed method for extracting object dimensions has an error rate of 6.492 on average and it runs at 250 frames per second, therefore it is also suitable for applications in automated warehouses.

Item Type:	Conference Paper
Publication:	Lecture Notes in Mechanical Engineering
Publisher:	Springer Science and Business Media Deutschland GmbH
Additional Information:	The copyright for this article belongs to the Springer Science and Business Media Deutschland GmbH.
Keywords:	Computer vision; Convolutional neural networks; Deep learning; Object classification; Palletization; ResNet-50; Warehouse automation
Department/Centre:	Division of Mechanical Sciences > Centre for Product Design & Manufacturing
Date Deposited:	05 Jul 2022 12:05
Last Modified:	13 Jul 2022 05:10
URI:	https://eprints.iisc.ac.in/id/eprint/74222

Actions (login required)

View Item