Moving Scenes Detection Based On DeblurGAN-v2 And YOLOv5

With the rapid growth in wood demand, the need for accurate and efficient counting during wood transportation has become increasingly prominent, especially in mobile scenarios such as transport vehicles and docks. Traditional manual counting methods are inefficient, prone to errors, and pose safety risks. To address the issues of low counting accuracy and efficiency caused by motion blur in images, this study proposes a wood counting method based on DeblurGAN-v2 and YOLOv5. First, industrial cameras are used to capture motion-blurred images of wood in mobile scenarios, and the images are labeled using the Labelimg tool to construct the original dataset. Next, DeblurGAN-v2 is utilized to deblur the original images, generating a clear deblurred dataset. Then, this deblurred dataset is used to train the YOLOv5 model for wood object detection and counting. Experimental results show that this method significantly improves the accuracy and efficiency of wood counting in mobile scenarios and can effectively tackle the challenges posed by motion blur, demonstrating good practical application potential.