DiCENet: Dimension-wise Convolutions for Efficient Networks
This work addresses the need for more efficient and high-performing neural networks for computer vision tasks, particularly on resource-constrained devices, and is incremental as it builds on existing separable convolution methods.
The paper tackles the problem of improving efficiency and performance in convolutional neural networks by introducing a novel convolutional unit called DiCE, which uses dimension-wise convolutions and fusion to encode spatial and channel information; the result is DiCENet, which achieves 2-4% higher accuracy on ImageNet compared to state-of-the-art models like MobileNetv2 and shows better generalization in tasks such as object detection.
We introduce a novel and generic convolutional unit, DiCE unit, that is built using dimension-wise convolutions and dimension-wise fusion. The dimension-wise convolutions apply light-weight convolutional filtering across each dimension of the input tensor while dimension-wise fusion efficiently combines these dimension-wise representations; allowing the DiCE unit to efficiently encode spatial and channel-wise information contained in the input tensor. The DiCE unit is simple and can be seamlessly integrated with any architecture to improve its efficiency and performance. Compared to depth-wise separable convolutions, the DiCE unit shows significant improvements across different architectures. When DiCE units are stacked to build the DiCENet model, we observe significant improvements over state-of-the-art models across various computer vision tasks including image classification, object detection, and semantic segmentation. On the ImageNet dataset, the DiCENet delivers 2-4% higher accuracy than state-of-the-art manually designed models (e.g., MobileNetv2 and ShuffleNetv2). Also, DiCENet generalizes better to tasks (e.g., object detection) that are often used in resource-constrained devices in comparison to state-of-the-art separable convolution-based efficient networks, including neural search-based methods (e.g., MobileNetv3 and MixNet. Our source code in PyTorch is open-source and is available at https://github.com/sacmehta/EdgeNets/