Comparison of Object Detection Technique for Applications in ETAT Automation Training Set
Karin Kosiyanurak, Pongpat Singsri, Soontaree Seangsri, Sommart Promput · 2023
Object detection is a type of artificial intelligence (AI) that is currently being developed and used widely. It can classify and detect objects in images or videos. This research aims to investigate object detection that is suitable and accurate for detecting the color of the workpieces in the training set. This study will experiment comparison of three object detection models: 1. OpenCV Python code via the Thonny software. Using the BGR color separation principle, the range of color values must be specified so that the model can separate different colors. 2. V4-tiny and 3. V7-tiny are models that use model training principles in the CiRA-CORE Platform. In this study, 1,271 images of workpieces processed with the Augmentation Technique were utilized to train the model. and 240 images for testing the model. To analyze of find a suitable and accurate model for detecting the color of workpieces Because light is important in visual or object detection in real-time this study divided the experiment into two conditions: LED Lighting and without LED Lighting. The results showed that V7-tiny is an object detection model that can detect the color of the target in both lighting conditions. The V7-tiny model has an accuracy of 97.92%.