Guiding Long-Short Term Memory for Image Caption Generation
arXiv:1509.04942
Abstract
In this work we focus on the problem of image caption generation. We propose an extension of the long short term memory (LSTM) model, which we coin gLSTM for short. In particular, we add semantic information extracted from the image as extra input to each unit of the LSTM block, with the aim of guiding the model towards solutions that are more tightly coupled to the image content. Additionally, we explore different length normalization strategies for beam search in order to prevent from favoring short sentences. On various benchmark datasets such as Flickr8K, Flickr30K and MS COCO, we obtain results that are on par with or even outperform the current state-of-the-art.
accepted by ICCV 2015
References in corpus (6)
- Sequence to Sequence Learning with Neural Networks
- Show, Attend and Tell: Neural Image Caption Generation with Visual Attention
- LSTM: A Search Space Odyssey
- Sequence Transduction with Recurrent Neural Networks
- Deep Fragment Embeddings for Bidirectional Image Sentence Mapping
- MatConvNet - Convolutional Neural Networks for MATLAB
Cited by in corpus (12)
- Generating Visual Explanations
- What value do explicit high level concepts have in vision to language problems?
- Semantic Compositional Networks for Visual Captioning
- An Effective Automatic Image Annotation Model Via Attention Model and Data Equilibrium
- Unpaired Image Captioning by Language Pivoting
- Storytelling of Photo Stream with Bidirectional Multi-thread Recurrent Neural Network
- Figure Captioning with Reasoning and Sequence-Level Training
- Image Captioning Based on a Hierarchical Attention Mechanism and Policy Gradient Optimization
- Unpaired Cross-lingual Image Caption Generation with Self-Supervised Rewards
- An Empirical Study of Language CNN for Image Captioning
- Hierarchical Memory Decoding for Video Captioning
- Image Captioning based on Deep Reinforcement Learning