Air quality estimation from sequential surveillance images using a unified CNN-RNN framework.

Viewed holistically, these figures reveal a clear trend: across CNN backbones, performance tends to follow the order VGG16 > ResNet50 > InceptionV3, and across RNN variants, accuracy improves from CNN-only to SimpleRNN to GRU to LSTM, which further highlights the central role of temporal modeling in this framework.

also in 12,136 other papers