Superpixel Image Classification using GNNs. • Used SLIC to segment superpixels on MNIST and CIFAR-10 images. • Trained different models using GCN, GAT, GraphSage, GIN, Hierarchical GNN and then compared the results for different number of node settings. Achieved the best accuracy for Hierarchical GNN with 200 nodes .