z-zheng / freenet

FPGA: Fast Patch-Free Global Learning Framework for Fully End-to-End Hyperspectral Image Classification (TGRS 2020) https://ieeexplore.ieee.org/document/9007624

Parsing Our Earth Using CV and RS (114.3459, 30.5347)

freenet's Issues

Some issues about the code.

I try to run your code on the Pavia data set.

Traceback (most recent call last):
File "d:/FreeNet-master/train.py", line 63, in
opts=args.opts)
File "D:\software\anaconda\lib\site-packages\simplecv\dp_train.py", line 44, in run
traindata_loader = make_dataloader(cfg['data']['train'])
File "D:\software\anaconda\lib\site-packages\simplecv\data\data_loader.py", line 9, in make_dataloader
raise ValueError('{} is not support now.'.format(dataloader_type))
ValueError: NewPaviaLoader is not support now.

Looking forward to hearing from you!

在安装simpleCV时出现报错

您好，我在终端安装simpleCV时出现了以下报错:
ERROR: Could not find a version that satisfies the requirement tensorboardX==1.7 (from simplecv) (from versions: none)
ERROR: No matching distribution found for tensorboardX==1.7
请问您知道如何解决吗，simpleCV是基于tensorflow框架的吗，但freenet好像是基于pytorch，很抱歉打扰您

About using one conv layer twice

FreeNet/module/freenet.py

Line 95 in 879f3de

out = self.fuse_3x3convs[i](inner)

Thanks for your great work!
I wonder why you self.fuse_3x3convs[0]() twice.

关于可视化的问题

作者你好，我是该领域的新手，感觉您论文写的超级棒，目前正在复现您的代码，2013_IEEE_GRSS_DF_Contest_CASI.tif也成功获取，且模型在三个数据集上的表现与您论文所述几乎一致，但作为一个跨领域的小白，我无法复现出您在论文中的那种图，比如Fig. 12. Visualization of the classification maps for the CASI Universityof Houston data set. (a) SVM. (b) S-CNN. (c) Gabor-CNN. (d) DFFN.(e) 3-D-GAN. (f) FreeNet. 那种可以给CASI数据集中15种城市土地覆盖类型涂上不同颜色，
还有fig 11那种的可视化Fig. 11. CASI University of Houston data set. (a) Color composite representation of the hyperspectral data using bands of 70, 50, and 20,as red, green, and blue, respectively. (b) Training samples. (c) Test samples.(d) Legend.以及fig10中的可视化，请问能麻烦您给指点一下思路或者有相关可以参考的源码吗？不胜感激（抱拳）

运行 train.py时报错

Traceback (most recent call last):
File "D:/论文代码书/代码/高光谱影像分类/全卷积FCN/FreeNet-master/train.py", line 60, in
train.run(config_path=args.config_path,
File "D:\Anaconda\envs\Pytorch\lib\site-packages\simplecv-0.3.4-py3.8.egg\simplecv\dp_train.py", line 29, in run
cfg = config.import_config(config_path)
File "D:\Anaconda\envs\Pytorch\lib\site-packages\simplecv-0.3.4-py3.8.egg\simplecv\util\config.py", line 5, in import_config
m = importlib.import_module(name='{}.{}'.format(prefix, config_name))
File "D:\Anaconda\envs\Pytorch\lib\importlib_init_.py", line 127, in import_module
return _bootstrap._gcd_import(name[level:], package, level)
File "", line 1014, in _gcd_import
File "", line 991, in _find_and_load
File "", line 973, in _find_and_load_unlocked
ModuleNotFoundError: No module named 'configs.None'

Test.py module

Hello!

I really appreciate your paper and sharing the code for it. I wonder is there an option to make a test on the trainned network on another image?
I saw test dict in config file, but I'm not sure it is implemented for now. Is there any plans for it or will you please suggest how can it be done better?

Thanks!