Performance analysis of convolutional neural networks for recognition task of similar industrial machining parts

Fahar, Muhammad Fikri (2022) Performance analysis of convolutional neural networks for recognition task of similar industrial machining parts. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full Text) Performance analysis of convolutional neural networks for recognition task of similar industrial machining parts.pdf - Submitted Version Download (1MB)

Abstract

With recent advances in convolutional neural network (CNN), there is an increased interest in applying this technology to industrial image processing. Specifically for automatic recognition of similar industrial machining parts. Misclassifying parts is one of the major issues faced by the operator in small and medium enterprises (SMEs) manufacturing industry. The manual inspection currently employed leaves room for human error due to inability to distinguish similar parts. As a result, incorrect machining parts could be sent to the costumer, decreasing the company’s reputation in the eye of their existing and potential customers. To overcome the traceability issue of similar machining parts, it is critical to incorporate automated and digital inspection. Therefore, the current work investigates the performance of different CNN models that can be integrated into a machine-vision system to perform automatic recognition tasks.

Actions (login required)

View Item