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 speed-up 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, as well as improved detection accuracy.

翻译：在数百层的所有地貌地图上,常规的深层电动神经网络(CNNs)在空间中统一应用各种变异操作器,这给实时应用带来了很高的计算成本。对于物体探测和语义分解等许多问题,我们可以从先验问题知识或低解析分解网络获得低成本的计算面罩。我们表明,这种计算面罩可以用来减少高分辨率主网络的计算量。曾经就小型任务探索过微弱激活CNN的变异功能,在物体分类精确度方面没有出现退化,但在理论FLOP方面却经常衡量收益,但与高度优化的密集变异实施相比,没有实现实际的加速。在这项工作中,我们利用计算面罩的宽度结构,提出了一种新型的基于节流的稀有变异算法算法。我们核实了我们分散的基于3DAR3D对象探测的光学CNNCN的有效性,并且我们报告了与密集变异性相比重大的墙时速增速率,以及探测准确性也提高了。