Browsing by Author "1813326642"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Open Access
    Destruction Detection Using Proprietary Neural Network
    (North South University, 2023) Abid Hasan Saheel; Fahim Hossain; Jannatul Ferdous Sristy; Dr. Atiqur Rahman (AQU); 1912084642; 1813326642; 1931533042
    Recently, there has been a boom in Machine learning and deep learning, and rightfully so. Because they have revolutionized the way automation works in any field. And having specialized object detection can automize and make the previously known undoable tasks possible but also make it relatively easy. And as the name suggests, detecting instances of semantic objects of a specific class (such as people, buildings, or cars) in digital photos and videos is the task of object detection. A branch of computer science linked to computer vision and image processing. And using specialized Object recognition, which is a more intensive object detection system. We used it to spot damage on vehicles through image processing. And now our project can correctly classify if a car is damaged or Not Damaged. We used a sequential neural network to train our model. Which went through 120 epochs and achieved an accuracy of 0.98. It also obtained precision, recall, and F1 scores of 0.83, 0.71, and 0.79, respectively. When a prediction value is obtained. It is based on a threshold of 0.5, the image is classified as "Car Damage" or "Car Not Damage." With this result, we can implement the model into any detection system, and it will detect a car and its damage value precisely, even with newer images. Keywords: Object Detection, Machine learning, Deep Learning, Neural Network, Damage detection, Vehicle damage.