opconty/pytorch_ctpn
This is a pytorch implementation of CTPN(Detecting Text in Natural Image with Connectionist Text Proposal Network). You may want to finetune from: https://drive.google.com/open?id=1JHhI4sEIXfs5gDa1I9AgJBY477HTzAd0
Python
Issues
- 13
vs2019 调试五轮,ep05_0.0278_0.0990_0.1268 就很难下降了
#31 opened by hsome2020 - 6
作者您好!请问下可否提供下训练好的weights?期待您的回复~
#33 opened by cay125 - 1
为什么您在model的forward函数把输入的channel first和channel last来回倒?一直channel first不行吗?
#7 opened by HandsomeBrotherShuaiLi - 8
训练时候12g显存爆了..
#16 opened by fanjianing - 1
training-data sample
#8 opened by mrocr - 1
requirements.txt & environment.yml
#11 opened by mrocr - 5
pretrained weights
#12 opened by wangshuai66666 - 1
请问一下 数据标签是什么呢 就是用lebelimg框出的字符区域吗
#43 opened by forbetteryou - 1
请问能提供一下训练好的权重文件吗?
#37 opened by ulanbator - 4
转成onnx和mnn
#42 opened by dengfenglai321 - 2
你们的batch_size 都是1吗?
#41 opened by jasnei - 0
为什么没有side-refinement分枝?
#40 opened by rolyeMMlin - 2
有关损失函数问题
#35 opened by wangguanhua - 0
- 0
ctpn_train.py: error: argument --num-workers: invalid int value: 'num_workers'
#38 opened by chenzhe222 - 3
- 0
请问一下,有人遇到过这个问题吗,网上说是标签的越界,是在损失函数里面报的错。
#36 opened by zhangts20 - 0
VOC2007_text_detection是哪一个数据集啊
#34 opened by bxhandhxb - 8
请问使用的数据集图片尺寸是多少呢,predict失败,无文本框
#32 opened by gxxie - 5
请问一下显存炸了怎么办
#30 opened by gxxie - 1
voc2007复现失败
#29 opened by xmyhhh - 2
性能如何?
#2 opened by Bachelorwangwei - 6
是否可以选择开源协议以利于共享
#27 opened by essie-chiang - 4
求助大大
#28 opened by kaixinbaba - 1
哪里有数据集
#26 opened by beijing-penguin - 1
怎么准备数据集呢?大神可以介绍一下嘛?
#20 opened by isyanan1024 - 2
Reason for loss does not falling
#15 opened by 980044579 - 5
- 1
- 1
PyTorch用哪个版本?
#25 opened by ZhangXinNan - 4
loss=0.44 on the SROIE dataset
#22 opened - 16
- 3
能再分享一下你训练好的模型吗
#4 opened by Sierkinhane - 0
android ocr
#19 opened by dssone - 1
number of epochs?
#18 opened by GreenGarnets - 0
数据集
#14 opened by zhouguanghui001 - 0
数据集连接失效了,能重新发一下吗
#13 opened by zhouguanghui001 - 1
pretrained weights is unavailable
#9 opened by ZhangWen0301 - 1
Push push your requirements.txt
#10 opened by elcolie - 2
the dropbox link is invalid now
#5 opened by weixinran - 0
- 2