DyNet: Dynamic Convolution for Accelerating Convolutional Neural Networks

Abstract

Convolution operator is the core of convolutional neural networks (CNNs) and occupies the most computation cost. To make CNNs more efficient, many methods have been proposed to either design lightweight networks or compress models. Although some efficient network structures have been proposed, such as MobileNet or ShuffleNet, we find that there still exists redundant information between convolution kernels. To address this issue, we propose a novel dynamic convolution method to adaptively generate convolution kernels based on image contents. To demonstrate the effectiveness, we apply dynamic convolution on multiple state-of-the-art CNNs. On one hand, we can reduce the computation cost remarkably while maintaining the performance. For ShuffleNetV2/MobileNetV2/ResNet18/ResNet50, DyNet can reduce 37.0/54.7/67.2/71.3% FLOPs without loss of accuracy. On the other hand, the performance can be largely boosted if the computation cost is maintained. Based on the architecture MobileNetV3-Small/Large, DyNet achieves 70.3/77.1% Top-1 accuracy on ImageNet with an improvement of 2.9/1.9%. To verify the scalability, we also apply DyNet on segmentation task, the results show that DyNet can reduce 69.3% FLOPs while maintaining Mean IoU on segmentation task.

References in corpus (16)

• Very Deep Convolutional Networks for Large-Scale Image Recognition
• Distilling the Knowledge in a Neural Network
• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• Caffe: Convolutional Architecture for Fast Feature Embedding
• FitNets: Hints for Thin Deep Nets
• Do Deep Nets Really Need to be Deep?
• Trained Ternary Quantization
• Compressing Neural Networks with the Hashing Trick
• Speeding up Convolutional Neural Networks with Low Rank Expansions
• Speeding-up Convolutional Neural Networks Using Fine-tuned CP-Decomposition
• CondConv: Conditionally Parameterized Convolutions for Efficient Inference
• Dual Attention Network for Scene Segmentation
• MetaPruning: Meta Learning for Automatic Neural Network Channel Pruning
• Good Subnetworks Provably Exist: Pruning via Greedy Forward Selection
• IGCV3: Interleaved Low-Rank Group Convolutions for Efficient Deep Neural Networks
• Dynamic Region-Aware Convolution

Cited by in corpus (8)

• Lightweight Salient Object Detection in Optical Remote-Sensing Images via Semantic Matching and Edge Alignment
• Computational Optics for Mobile Terminals in Mass Production
• Learning A Single Network for Scale-Arbitrary Super-Resolution
• Collaboration of Experts: Achieving 80% Top-1 Accuracy on ImageNet with 100M FLOPs
• LAConv: Local Adaptive Convolution for Image Fusion
• Instance and Pair-Aware Dynamic Networks for Re-Identification
• Decoupled Dynamic Filter Networks
• Towards Understanding the Effectiveness of Attention Mechanism