search

yelusaleng / RRU-Net

Official repository for "RRU-Net: The Ringed Residual U-Net for Image Splicing Forgery Detection" (CVPRW 2019)
107 stars 18 forks source link

数据集处理 #34

Open lzJune opened 1 year ago

lzJune commented 1 year ago

您好，我在看您的代码的train.py中有以下这段 dir_img = './data/data_{}/train/tam/'.format(dataset) dir_mask = './data/data_{}/train/mask/'.format(dataset) dir_logs = './result/logs/{}/{}/'.format(dataset, model) 但是我在下载CASIA数据集后没发现有tam、mask之类的子目录，请问是您自己构建的吗，我如果想要重新训练您的代码，该如何对我的数据集进行操作呢，CASIA2.0数据集如下图，请指教~ QQ截图20230320112927 QQ截图20230320113002

yelusaleng commented 1 year ago

tam就是放篡改图像的目录，mask放对应的GT图像

lzJune commented 1 year ago

谢谢，请问tam中是只需要放篡改图像吗，那CASIA官方给的数据集中Au文件夹下的真实数据是有什么作用吗 GIFKVW LDP4` R IC)M2X5T

yelusaleng commented 1 year ago

是。真实数据暂时用不到

lzJune commented 1 year ago

好的，谢谢~

chaukiching commented 1 year ago

好的，谢谢~

您好，我想问一下要如何进行训练和测试，我在各子目录里放完对应的图像后，运行train.py后报错是什么原因。 line 67, in train_net imgs = np.array([i[0] for i in b]).astype(np.float32) ValueError: setting an array element with a sequence. The requested array has an inhomogeneous shape after 2 dimensions. The detected shape was (6, 3) + inhomogeneous part.