2018 · 202 citations · 32 references
ComputingComputer EngineeringComputer ScienceComputer VisionMachine LearningMachine VisionDeep LearningHardware AccelerationObject DetectionObject RecognitionPattern RecognitionSemantic Segmentation3D Semantic SegmentationAccuracy-computational Cost TradeoffConvolutionObject ClassificationSparse Network InferenceSparse Network LearningSparsity Problem
Conventional deep convolutional neural networks (CNNs) apply convolution operators uniformly in space across all feature maps for hundreds of layers - this incurs a high computational cost for real-time applications. For many problems such as object detection and semantic segmentation, we are able to obtain a low-cost computation mask, either from a priori problem knowledge, or from a low-resolution segmentation network. We show that such computation masks can be used to reduce computation in the high-resolution main network. Variants of sparse activation CNNs have previously been explored on small-scale tasks and showed no degradation in terms of object classification accuracy, but often measured gains in terms of theoretical FLOPs without realizing a practical speedup when compared to highly optimized dense convolution implementations. In this work, we leverage the sparsity structure of computation masks and propose a novel tiling-based sparse convolution algorithm. We verified the effectiveness of our sparse CNN on LiDAR-based 3D object detection, and we report significant wall-clock speed-ups compared to dense convolution without noticeable loss of accuracy.
32
Deep Residual Learning for Image Recognition
Kaiming He, Xiangyu Zhang, Shaoqing Ren et al. · 2016 · 222.1K citations · Full text
Focal Loss for Dense Object Detection
Tsung-Yi Lin, Priya Goyal, Ross Girshick et al. · 2017 · 25.7K citations
Hengshuang Zhao, Jianping Shi, Xiaojuan Qi et al. · 2017 · 15.5K citations