Research map / Vision-language and multimodal / Human and embodied modeling
Reference Based LSTM for Image Captioning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| weight words in the training loss by relevance to the image (ablation) | VGG-16 + LSTM · MS COCO (no finetuning) | CIDEr: 70.2 -> 71.8 | Helped |
| weight words in the training loss by relevance to the image (ablation) | VGG-16 + LSTM finetuned · MS COCO | CIDEr: 92.8 -> 94.0 | Helped |
| weight words in the training loss (ablation) | VGG-16 + LSTM finetuned · MS COCO | BLEU-4: 30.1 -> 30.4 | Helped |
| weight words in the training loss (ablation) | VGG-16 + LSTM · MS COCO (no finetuning) | BLEU-1: 65.0 -> 65.1 | No clear effect |
| add consensus score of neighbor-image captions to beam search (alpha from 0 to a moderate value) (ablation) | R-LSTM · MS COCO | CIDEr | Helped |
| rely too heavily on references (large alpha) (ablation) | R-LSTM · MS COCO | CIDEr | Hurt |
| use different alpha at early vs final pool stage (alpha2=0.4, alpha1=0.2) (ablation) | R-LSTM · MS COCO | CIDEr | Helped |
| use full R-LSTM vs strongest prior results | R-LSTM · MS COCO Karpathy split | CIDEr: 93.8 -> 102.9 | Helped |
Closest papers by meaning
- Convolutional Image Captioning (CVPR 2018 · similarity 0.91)
- Incorporating Copying Mechanism in Image Captioning for Learning Novel Objects (CVPR 2017 · similarity 0.90)
- Pointing Novel Objects in Image Captioning (CVPR 2019 · similarity 0.90)
- Video Captioning With Transferred Semantic Attributes (CVPR 2017 · similarity 0.88)
- Noise-Aware Image Captioning with Progressively Exploring Mismatched Words (AAAI 2024 · similarity 0.85)
- BalCapRL : A Balanced Framework for RL-Based MLLM Image Captioning (NeurIPS 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.