This is the official repo of dataset Language-prompted Localization Dataset proposed in Griffon, accepted by ECCV2024.

Paper or resources for more information: [Paper][Code]

Dataset Details:

• template folder: The instruction prompts generated by ChatGPT(driven by GPT-4V). For the pretraining stage, you can use 1v1.json for REC and NvN.json for Object Detection.
• Pre-training Data: Download the Visual Genome, MSCOCO train2014 & train 2017, and Objects365-2023 images from official repo. We provide the converted annotations of Visual Genome and RefCOCO series following Shikra and the filtered Object365 annotations. The raw annotation file of MSCOCO train2017 can be directly used.
• Instruction Data: Additionally downloading the flickr30K Entities images for the 1vN data.

License: Attribution-NonCommercial 4.0 International It should abide by the policy of OpenAI.

Citation

@misc{zhan2023griffon,
      title={Griffon: Spelling out All Object Locations at Any Granularity with Large Language Models}, 
      author={Yufei Zhan and Yousong Zhu and Zhiyang Chen and Fan Yang and Ming Tang and Jinqiao Wang},
      year={2023},
      eprint={2311.14552},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}