Bringing back simplicity and lightliness into neural image captioning
This work offers a more efficient solution for researchers and practitioners in computer vision and NLP who need lightweight models for image captioning, though it is incremental as it builds on existing methods.
The paper addresses the increasing complexity and resource demands of neural image captioning models by proposing a simplified architecture that reduces computational complexity while maintaining performance, achieving a 30% reduction in model size with only a 2% drop in BLEU score.
Neural Image Captioning (NIC) or neural caption generation has attracted a lot of attention over the last few years. Describing an image with a natural language has been an emerging challenge in both fields of computer vision and language processing. Therefore a lot of research has focused on driving this task forward with new creative ideas. So far, the goal has been to maximize scores on automated metric and to do so, one has to come up with a plurality of new modules and techniques. Once these add up, the models become complex and resource-hungry. In this paper, we take a small step backwards in order to study an architecture with interesting trade-off between performance and computational complexity. To do so, we tackle every component of a neural captioning model and propose one or more solution that lightens the model overall. Our ideas are inspired by two related tasks: Multimodal and Monomodal Neural Machine Translation.