Defect Detection in Manufacturing

• Author(s): Rohit Pendse ; Harshal Rajput ; Shubham Saraf ; Atharva Sarwate ; Jyoti Jadhav
• Paper ID: 1704201
• Page: 272-275
• Published Date: 28-03-2023
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 6 Issue 9 March-2023

Abstract

The detection of quality defects is crucial in quality control in industrial production. An enhanced manufacturing defect detection method based on the You Only Look Once (YOLO) model is proposed in order to address the issues of ineffective detection brought on by conventional human inspection and indistinct features in industrial defect detection. Due to their complexity and distinctive characteristics, flaws might be difficult to detect in produced parts traveling on conveyor belts. These flaws are unique to each production line and can be found on many of them. YOLOV7 is real-time object detection technique, which internally utilizes a convolutional neural network (CNN), is utilized to detect these inaccuracies. Despite the tiny dataset and minimal network adjustments, YOLOV7 can obtain a mean average precision (mAP) of above 70%. The network can get the bounding box coordinates that are detected and that correspond to one of the classes in the annotated data.

Keywords

YOLO, SSD, Convolutional Neural Network, Object Detection.

Citations

IRE Journals:
Rohit Pendse , Harshal Rajput , Shubham Saraf , Atharva Sarwate , Jyoti Jadhav "Defect Detection in Manufacturing" Iconic Research And Engineering Journals Volume 6 Issue 9 2023 Page 272-275

IEEE:
Rohit Pendse , Harshal Rajput , Shubham Saraf , Atharva Sarwate , Jyoti Jadhav "Defect Detection in Manufacturing" Iconic Research And Engineering Journals, 6(9)