Table 1 Summary of the used topic, image, and language models in the topic-based approaches.
From: Novel concept-based image captioning models using LSTM and multi-encoder transformer architecture
Topic modeling technique | Image encoder | Language model | |
|---|---|---|---|
Topic-based captioning7 | LDA | Inception-V3 | LSTM |
Show and tell more22 | LDA | ResNet-152 | Two layers of LSTM |
What topics do images say23 | LDA | VGGNet | LSTM |
Topic-oriented (NeuralTalk2-T-oe)28 | LDA & NMF | VGGNet | LSTM |
Topic-guided attention (VA)8 | LDA | VGGNet | LSTM with attention |
Topic-sensitive30 | HDP | AlexNet | LSTM |
