Empirical Autopsy of Deep Video Captioning Frameworks
This work addresses a gap in video captioning research by providing empirical guidance for component selection, which is incremental but useful for researchers in this domain.
The paper tackles the lack of systematic investigation into component selection in deep video captioning frameworks, finding that careful selection of components like CNNs and feature transformations can lead to substantial performance gains on the MSVD dataset without major pipeline changes.
Contemporary deep learning based video captioning follows encoder-decoder framework. In encoder, visual features are extracted with 2D/3D Convolutional Neural Networks (CNNs) and a transformed version of those features is passed to the decoder. The decoder uses word embeddings and a language model to map visual features to natural language captions. Due to its composite nature, the encoder-decoder pipeline provides the freedom of multiple choices for each of its components, e.g the choices of CNNs models, feature transformations, word embeddings, and language models etc. Component selection can have drastic effects on the overall video captioning performance. However, current literature is void of any systematic investigation in this regard. This article fills this gap by providing the first thorough empirical analysis of the role that each major component plays in a contemporary video captioning pipeline. We perform extensive experiments by varying the constituent components of the video captioning framework, and quantify the performance gains that are possible by mere component selection. We use the popular MSVD dataset as the test-bed, and demonstrate that substantial performance gains are possible by careful selection of the constituent components without major changes to the pipeline itself. These results are expected to provide guiding principles for future research in the fast growing direction of video captioning.