ZJ

what is the dataset of pre-training model？ could anyone offer a url about datast

> when i use Muti GPU,test fuction val raise > assert t.numel() == length.sum(), "texts with length: {} does not match declared length: {}".format(t.numel(), length.sum()) > AssertionError: texts with length:...

Do you create your own datasets? Does the author provide open source datasets ?

Do you create your own datasets? Does the author provide open source datasets ?

> What's your test set? I try to reproduce the results of CRNN, but got really bad performance. > I create syth90k_train (about 7.2M images) and syth90k_val (about 0.8M images)...

Do you rembmber the map_size in [https://github.com/bgshih/crnn/blob/master/tool/create_dataset.py#L37](https://github.com/bgshih/crnn/blob/master/tool/create_dataset.py#L37) when make train lmdb, that includes 70K images. I set the map_size as 20G, but not enough

Thanks. did you finally reproduce the results of the author's model? Now， I am retraining the model with a simplified network. Could you tell me the batchsize you set when...

Do you get the SynthText Datasets? could you share it for me

train_320时候，[随机裁剪](https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB/blob/96f430f4e2ce376752baa949528e50290808cdb0/vision/transforms/transforms.py#L367)这里没有问题吗？虽然裁剪的是1:1，但是最后min_iou都是会小于1的吧，调用[RandomSampleCrop_v2](https://github.com/Linzaer/Ultra-Light-Fast-Generic-Face-Detector-1MB/blob/96f430f4e2ce376752baa949528e50290808cdb0/vision/transforms/transforms.py#L352)按道理还是会直接原图resize 320x240的啊。我理解的有误吗？ @eternalgogi92 @Linzaer

弄明白了，没问题，之前只看到了函数注释，没看源码，注释虽然是说IOU，实际计算不是的，意思是如果没有包含完成的gt box crop的就不要