فائزه مهدي ملكي

Object detection is a key area in image processing and computer vision that involves identifying the spatial location and classification of visual objects from specific classes in digital images. This technology has widespread applications in fields such as remote sensing, robotic vision, video surveillance, and various industries. In this thesis, object detection in images is examined using convolutional neural networks. These networks have the capability to automatically extract features from training samples without the need for manual feature definition and extraction. The primary objective of this study is to detect and analyze the size distribution of objects in industrial images. To achieve this, a dedicated dataset comprising images containing the target objects was collected through imaging in a laboratory environment. After performing preprocessing and applying appropriate data augmentation techniques, this dataset was used for training and eva‎luating the YOLOv8 model based on convolutional neural networks. The results of this study indicate that the proposed model, after the learning process, can accurately detect all objects in the images with a mean average precision (MAP)of 99%. Moreover, the obtained values for other model performance eva‎luation metrics demonstrate its high precision and speed in the object detection process.

تشخيص اشياء , يادگيري عميق , شبكه عصبي , شبكه عصبي كانولوشن , مدل يولو

Object Detection , Deep Learning , Neural Network , Convolutional Neural Network , YOLO Model

Faezeh Mahdimaleki

Dr. Mohammadreza Jafarfard ,Dr. Babak Zare