Semantic segmentation methods have made impressive progress with deep learning. However, while achieving higher and higher accuracy, state-of-the-art neural networks overlook the complexity of architectures, which typically feature dozens of millions of trainable parameters. Consequently, these networks requires high computational ressources and are mostly not suited to perform on edge devices with tight resource constraints, such as phones, drones, or satellites. In this work, we propose two highlycompact neural network architectures for semantic segmentation of images, which are up to 100 000 times less complex than state-of-the-art architectures while approaching their accuracy. To decrease the complexity of existing networks, our main ideas consist in exploiting lightweight encoders and decoders with depth-wise separable convolutions and decreasing memory usage with the removal of skip connections between encoder and decoder. Our architectures are designed to be implemented on a basic FPGA such as the one featured on the Intel Altera Cyclone V family of SoCs. We demonstrate the potential of our solutions in the case of binary segmentation of remote sensing images, in particular for extracting clouds and trees from RGB satellite images.
Graph Neural Networks (GNNs) have emerged as a powerful and flexible framework for representation learning on irregular data. As they generalize the operations of classical CNNs on grids to arbitrary topologies, GNNs also bring much of the implementation challenges of their Euclidean counterparts. Model size, memory footprint, and energy consumption are common concerns for many realworld applications. Network binarization allocates a single bit to network parameters and activations, thus dramatically reducing the memory requirements (up to 32x compared to single-precision floating-point parameters) and maximizing the benefits of fast SIMD instructions of modern hardware for measurable speedups. However, in spite of the large body of work on binarization for classical CNNs, this area remains largely unexplored in geometric deep learning. In this paper, we present and evaluate different strategies for the binarization of graph neural networks. We show that through careful design of the models, and control of the training process, binary graph neural networks can be trained at only a moderate cost in accuracy on challenging benchmarks. In particular, we present the first dynamic graph neural network in Hamming space, able to leverage efficient k-NN search on binary vectors to speed-up the construction of the dynamic graph. We further verify that the binary models offer significant savings on embedded devices.
Automatic road graph extraction from aerial and satellite images is a long-standing challenge. Existing algorithms are either based on pixel-level segmentation followed by vectorization, or on iterative graph construction using next move prediction. Both of these strategies suffer from severe drawbacks, in particular high computing resources and incomplete outputs. By contrast, we propose a method that directly infers the final road graph in a single pass. The key idea consists in combining a Fully Convolutional Network in charge of locating points of interest such as intersections, dead ends and turns, and a Graph Neural Network which predicts links between these points. Such a strategy is more efficient than iterative methods and allows us to streamline the training process by removing the need for generation of starting locations while keeping the training end-to-end. We evaluate our method against existing works on the popular RoadTracer dataset and achieve competitive results. We also benchmark the speed of our method and show that it outperforms existing approaches. Our method opens the possibility of in-flight processing on embedded devices for applications such as real-time road network monitoring and alerts for disaster response.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.