Survey of Image Based Graph Neural Networks
This is an incremental survey paper that provides a classification framework and comparative analysis for researchers working on graph neural networks applied to image data.
This survey analyzes image-based graph neural networks by proposing a three-step classification approach that converts images to superpixels (reducing input data by 30%), generates region adjacency graphs, and uses graph convolutional networks for classification, finding that spectral-based models outperform spatial-based models and classical CNNs with lower computational cost.
In this survey paper, we analyze image based graph neural networks and propose a three-step classification approach. We first convert the image into superpixels using the Quickshift algorithm so as to reduce 30% of the input data. The superpixels are subsequently used to generate a region adjacency graph. Finally, the graph is passed through a state-of-art graph convolutional neural network to get classification scores. We also analyze the spatial and spectral convolution filtering techniques in graph neural networks. Spectral-based models perform better than spatial-based models and classical CNN with lesser compute cost.