Landslides are a widespread natural peril triggering remarkable destruction. The landslides induced by rainfall arise in all highland zones, triggering living beings and nature to be disarrayed. Global warming and ensuing changes in climate patterns have radically transformed the environment, resulting in a substantial increase in rainfall. Landslides and historical rainfall have a strong correlation. In today’s world of information and Communication Technologies (ICT), natural catastrophes can be regulated effectively. The study proposes a multi-modal feature fusion, based on an intermediate fusion approach wherein the features, respective to each modality, are fused before classification. Two diverse data modalities and their respective models are fitted to build a multi-modal model. Deep learning sequential and computer vision models, Long Short-term Memory and Bidirectional Long Short-Term Memory, based on the performance, are selected to build a multi-modal fusion model to predict rainfall-induced landslides. The multi-modal is trained and validated using rainfall time-series data and a landslide image dataset. The result obtained depicts that the fusion approach is effective in predicting rainfall-induced landslides.