PSENet.pytorch icon indicating copy to clipboard operation
PSENet.pytorch copied to clipboard

A pytorch re-implementation of PSENet: Shape Robust Text Detection with Progressive Scale Expansion Network

Results 26 PSENet.pytorch issues
Sort by recently updated
recently updated
newest added

toplayer有3层,模型文件里只有1层 ![image](https://user-images.githubusercontent.com/36028079/92208523-f7180e80-eebd-11ea-9be0-153515cc72dc.png)

得到的框普遍往左边有微弱偏移,严重的无法框住最右端,看了代码,不知道哪里出了问题 ,有遇到的么 ?

Traceback (most recent call last): File "train.py", line 263, in main() File "train.py", line 177, in main num_workers=int(config.workers)) File "/home/yian/anaconda3/envs/psenet_pytorch/lib/python3.7/site-packages/torch/utils/data/dataloader.py", line 176, in __init__ sampler = RandomSampler(dataset) File "/home/yian/anaconda3/envs/psenet_pytorch/lib/python3.7/site-packages/torch/utils/data/sampler.py", line...

在运行train.py进行训练时,当使用gpu时,运行到loss.backward时就会报出 Process finished with exit code -1073741819 (0xC0000005)。我是在pycharm中运行的代码,参数配置如下: 2020-06-09 17:24:46 INFO utils.py: logger init finished 2020-06-09 17:24:46 INFO train.py: {'Lambda': 0.7, 'OHEM_ratio': 3, 'backbone': 'resnet50', 'checkpoint': '', 'data_shape': 32, 'display_input_images':...

作者的代码很棒,但是这几天我在实际的实验中遇到了一些问题并进行了其他几处改进,在这里分享一下。 1. eval失败。解决:script.py文件中16,17行改为 'GT_SAMPLE_NAME_2_ID': 'gt_img_([0-9]+).txt', 'DET_SAMPLE_NAME_2_ID': 'res_img_([0-9]+).txt' 2. 其他python版本编译pse.so出错。我修改makefile文件编译成功了。详情见:https://blog.csdn.net/ab0902cd/article/details/88352417 3. 训练过程中writer保存了很多图片,但是我感觉用不到,而且保存图片后log文件会很大,可以添加一个if条件。例如添加: if config.display_input_images: 4. scale设置2或4时报错,尺寸不匹配。原因是模型得到的图是160x160,但是label图是640x640的。 解决:看了一下原作者的代码,train时不设置scale,test时设置scale。 ``` if self.train: x = F.interpolate(x, size=(H, W), mode='bilinear', align_corners=True) else: x = F.interpolate(x,...