ericosmic

Results 38 issues of ericosmic

I training my yolov4 for custom data based on alexeyab rep. I try convert weight to tf file by use save_model.py , but it give a error : ![image](https://user-images.githubusercontent.com/35327931/116660317-70bcb980-a9c5-11eb-9a4b-0257e65a794f.png) I...

我希望能将标注文件tmp_label.txt中的中文标注改为指向中文辞典的索引标注,可以在源代码中进行修改吗?

when I run the demo . it give this error, who can help me , please: ![image](https://user-images.githubusercontent.com/35327931/56257909-61c97e00-6100-11e9-9c5c-0d1fb6a92929.png)

when I build dependence by use npm install , it would produce error : No matching version found for tonal-chord@^1.1.4 , I found tonal-chord doesn't have version 1.1.4, so I...

因为crnn的训练集很大,所以我抽取了10张图片和对应的data_train.txt数据进行训练,只是想跑通训练过程,但是训练中报错如下: ![2019-05-14 11-04-38 的屏幕截图](https://user-images.githubusercontent.com/35327931/57667950-25723a80-7638-11e9-8a78-45793a229946.png) 有人知道需要错误原因吗

如题, 1 如果文本行间距比较小,ctpn的检测效果就会变得很差,会遗漏很多行的文本, 2 已经被检测到的文本行头尾也会出现遗漏一两个字的情况,不能将所有文字都框选到。 这两个问题是否需要经过重新训练来解决呢? 还是可以通过超参数调整来解决呢?

@YCG09 我使用自己生成的2000的小数据集(只包含汉字数字)来fine-tune 预训练模型,但是试了很多次训练的acc始终停留在0.23左右,loss~54,可以给一些建议吗?

I try to change path to /usr/bin/python3 , but it still give same error message. Have anyone met same problem?

I find Goto-definition function work well for my project in one of remote servers, but when I want to use it in other project in the other server, it didn't...

exit with code 9009, signal null