Image classification and object detection for low-resolution images and small objects with SPD-Conv
Image classification and object detection for low-resolution images and small objects with SPD-Conv
No More Strided Convolutions or Pooling: A New CNN Building Block for Low-Resolution Images and Small Objects
arXiv paper abstract https://arxiv.org/abs/2208.03641v1
arXiv PDF paper https://arxiv.org/pdf/2208.03641v1.pdf
GitHub https://github.com/labsaint/spd-conv
Convolutional neural networks (CNNs) … in …image classification and object detection … degrades rapidly on tougher tasks where images are of low resolution or objects are small.
… this roots in … the use of strided convolution and/or pooling layers, which results in a loss of fine-grained information and learning of less effective feature representations.
… propose a new CNN building block called SPD-Conv in place of each strided convolution layer and each pooling layer (thus eliminates them altogether).
SPD-Conv is comprised of a space-to-depth (SPD) layer followed by a non-strided convolution (Conv) layer, and can be applied in most if not all CNN architectures.
… create new CNN architectures by applying SPD-Conv to YOLOv5 and ResNet
… approach significantly outperforms state-of-the-art deep learning models … on … low-resolution images and small objects …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
