轻量级模型AC-Lite为阿萨姆语图像描述生成提供高效解决方案
AC-Lite : A Lightweight Image Captioning Model for Low-Resource Assamese Language
- 用轻量网络替代重型组件，降低计算开销
- 在COCO-AC数据集上达82.3的CIDEr得分，仅需2.45 GFLOPs
- 适合资源受限场景下的低资源语言图文生成
现有图像描述生成大多依赖计算密集型深度神经网络，且仅支持英语，难以跨语言和无障碍普及。本文提出AC-Lite，一种面向低资源阿萨姆语的轻量级图像描述模型。通过替换计算复杂的深层组件，AC-Lite显著降低计算需求。基于大量消融实验，选择ShuffleNetv2x1.5作为图像特征提取器，结合基于GRU的语言解码器与双线性注意力机制，在保持性能的同时实现最小计算开销。在COCO-AC数据集上，该模型取得82.3的CIDEr分数，仅需2.45 GFLOPs和22.87M参数。
原文摘要 · Abstract (English)
Most existing works in image caption synthesis use computation heavy deep neural networks and generates image descriptions in English language. This often restricts this important assistive tool for widespread use across language and accessibility barriers. This work presents AC-Lite, a computationally efficient model for image captioning in low-resource Assamese language. AC-Lite reduces computational requirements by replacing computation-heavy deep network components with lightweight alternatives. The AC-Lite model is designed through extensive ablation experiments with different image feature extractor networks and language decoders. A combination of ShuffleNetv2x1.5 with GRU based language decoder along with bilinear attention is found to provide the best performance with minimum compute. AC-Lite was observed to achieve an 82.3 CIDEr score on the COCO-AC dataset with 2.45 GFLOPs and 22.87M parameters.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。