-
I've noticed that sometimes image can be outside of source image, code below, you can test it. I've also make some quick fixes, I hope you know how to do it better
```python
data = [
(663, 2495…
-
你好,想问一下train.py里的imgW设置为256,是按照10位纯中文字符的长度来设置的吗?如果是中英文加数字的混合数字应该怎样设置imgW呢
-
Should I change imgH and imgW from their default values in order to train both vertical and horizontal text?
Should imgH and imgW have the same value?
I would be grateful if you could tell me the …
-
Hey, I don't know it this is OK or if this is the place to do something like that, but I was wondering if I could share a patch I did some time ago and maybe add it to the project.
I am no C progra…
-
The config file is as follows.
```yml
Global:
device: gpu
epoch_num: 100
log_smooth_window: 20
print_batch_step: 10
output_dir: ./output/rec/focalsvtr_smtr_ch_aug
save_epoch_step: …
-
I think it's quite a shame that you can't insert any other images except jpg's.
API wise: It should be fairly easy detecting what kind of image we have from the first few bytes of the image. If not t…
-
## Environment info
- Host: Linux
- xbvr docker container version: 0.4.24
- mariadb docker container version: 1:11.3.2+maria~ubu2204
- configured mariadb via the dockerfile in the xbvr wiki loca…
-
public static void Retina(NcnnDotNet.OpenCV.Mat src)
{
// Model from https://github.com/nihui/ncnn-assets/tree/master/models
Net net = new Net();
net.Opt.UseVulka…
-
看train.py文件的第98行代码:
image = torch.FloatTensor(opt.batchSize, 3, opt.imgH, opt.imgH)
按照常理来说,应该是image = torch.FloatTensor(opt.batchSize, 3, opt.imgH, opt.imgW)
可是我改成这样之后重新训练,不收敛的。请问作者,这里的image为什么要定义成…
-
[[JFImageManager sharedManager] thumbWithAsset:asset resultHandler:^(UIImage *result) {
//do something
}];
这个方法会返回图片两次 第一次返回略缩图 第二次返回 大图 如何判断是大图还是略缩图