StickerTAG
收藏资源简介:
StickerTAG是首个多标签贴纸数据集,由哈尔滨工业大学深圳分校等机构创建,包含461个标签和13,571对贴纸-标签注释,旨在深入理解贴纸的多重含义。数据集通过严格的标签构建和贴纸注释过程创建,涵盖了丰富的情感和对话词汇,反映了贴纸在实际对话中的多样性。该数据集适用于提升贴纸在多标签识别任务中的应用,特别是在需要精确区分贴纸细微含义的场景中。
StickerTAG is the first multi-label sticker dataset developed by Harbin Institute of Technology, Shenzhen (HIT Shenzhen) and other institutions. It consists of 461 labels and 13,571 sticker-label annotation pairs, aiming to provide in-depth insights into the multiple connotations of stickers. The dataset is constructed through rigorous label construction and sticker annotation procedures, covering abundant emotional and conversational vocabulary, which reflects the diversity of stickers in real-world conversations. This dataset is suitable for enhancing the application of stickers in multi-label recognition tasks, particularly in scenarios requiring precise differentiation of the subtle meanings of stickers.

- 1Towards Real-World Stickers Use: A New Dataset for Multi-Tag Sticker Recognition哈尔滨工业大学深圳分校 · 2024年



