FLODCAST: Flow and depth forecasting via multimodal recurrent architectures

Andrea Ciamarra, Federico Becattini, Lorenzo Seidenari, Alberto Del Bimbo

Abstract: Highlights•We propose FLODCAST, a joint future optical flow and depth estimator•FLODCAST mitigates error accumulation that typically limits autoregressive models•FLODCAST eliminates the need for different models with different prediction horizons.•We improve both modalities yielding state-of-the-art results.