Casting Visual Detection Based on improved Convolutional Network
For automatical diagnosing different defects in castings,a visual inspection system is designed to inspect the appearance of casting products.The multi-head self-attention module is used to enhance the downsampling process of the image recognition network,enabling the classic image classification network to obtain global feature information and promote the model's ability to identify small cracks and large-scale burrs on the appearance of castings.The experimental results show that the improved convolutional network has a higher recognition accuracy for casting defects,and can more accurately determine the type of casting defects.
万方数据
维普
