cuiziteng / Illumination-Adaptive-Transformer

[BMVC 2022] You Only Need 90K Parameters to Adapt Light: A Light Weight Transformer for Image Enhancement and Exposure Correction. SOTA for low light enhancement, 0.004 seconds try this for pre-processing.
Apache License 2.0
441 stars 43 forks source link

你好,我怎么样才能在最新版的mmdetection中使用dataset_type='ExdarkDataset'? #62

Closed Panhonganadd closed 3 months ago

Panhonganadd commented 8 months ago

我最近在学mmdetection,然后想用你们提供的exdark数据集训练模型,但是不知道你们是怎么定义的适合exdark的新的dataset_type

cuiziteng commented 7 months ago

数据集格式定义在 https://github.com/cuiziteng/Illumination-Adaptive-Transformer/tree/main/IAT_high/IAT_mmdetection/mmdet/datasets 下面的 https://github.com/cuiziteng/Illumination-Adaptive-Transformer/blob/main/IAT_high/IAT_mmdetection/mmdet/datasets/exdark.py 文件中,采用的VOC格式数据