UpResNet10のcaffemodelをwaifu2x-chainerのUpResNet10モデルから生成するスクリプト追加、生成したUpResNet10モデルとwaifu2x-chainerのUpResNet10モデルの出力を比較するスクリプト追加

appendix/check_diff_upresnet10.py

@@ -0,0 +1,63 @@
+import os
+import os.path as osp
+import sys
+import google.protobuf as pb
+from argparse import ArgumentParser
+import numpy as np
+import shutil
+
+import caffe
+from caffe.proto import caffe_pb2
+
+sys.path.append('waifu2x-chainer')
+from lib import srcnn
+import chainer
+
+
+def main():
+    caffe.set_mode_cpu()
+
+    model_name = 'UpResNet10'
+    model_dir = 'waifu2x-chainer/models/{}'.format(model_name.lower())
+    model_class = srcnn.archs[model_name]
+
+    for filename in os.listdir(model_dir):
+        basename, ext = os.path.splitext(filename)
+        if ext == '.npz':
+            model_path = os.path.join(model_dir, filename)
+            print(model_path)
+            channels = 3 if 'rgb' in filename else 1
+            model = model_class(channels)
+            chainer.serializers.load_npz(model_path, model)
+
+            model.to_cpu()
+
+            params = {}
+            for path, param in model.namedparams():
+                params[path] = param.array
+
+            net = caffe.Net('upresnet10_3.prototxt', caffe.TEST)
+            for key in net.params:
+                l = len(net.params[key])
+                net.params[key][0].data[...] = params[key + '/W']
+                if l >= 2:
+                    net.params[key][1].data[...] = params[key + '/b']
+
+            input_data = np.empty(net.blobs['input'].data.shape, dtype=np.float32)
+            input_data[...] = np.random.random_sample(net.blobs['input'].data.shape)
+
+            net.blobs['input'].data[...] = input_data
+            ret = net.forward()
+
+            input_data = np.empty(net.blobs['input'].data.shape, dtype=np.float32)
+            input_data[...] = np.random.random_sample(net.blobs['input'].data.shape)
+
+            net.blobs['input'].data[...] = input_data
+            ret = net.forward()
+
+            batch_y = model(input_data)
+            print(batch_y.array - ret['/conv_post'])
+
+if __name__ == '__main__':
+    caffe.init_log(3)
+    main()

appendix/gen_caffemodel_upresnet10.py

@@ -0,0 +1,61 @@
+import os
+import os.path as osp
+import sys
+import google.protobuf as pb
+from argparse import ArgumentParser
+import numpy as np
+import shutil
+
+import caffe
+from caffe.proto import caffe_pb2
+
+sys.path.append('waifu2x-chainer')
+from lib import srcnn
+import chainer
+
+fname_convert_table = {
+    'anime_style_noise0_scale_rgb': 'noise0_scale2.0x_model',
+    'anime_style_noise1_scale_rgb': 'noise1_scale2.0x_model',
+    'anime_style_noise2_scale_rgb': 'noise2_scale2.0x_model',
+    'anime_style_noise3_scale_rgb': 'noise3_scale2.0x_model',
+    'anime_style_scale_rgb': 'scale2.0x_model',
+}
+
+def main():
+    caffe.set_mode_cpu()
+
+    model_name = 'UpResNet10'
+    model_dir = 'waifu2x-chainer/models/{}'.format(model_name.lower())
+    model_class = srcnn.archs[model_name]
+
+    for filename in os.listdir(model_dir):
+        basename, ext = os.path.splitext(filename)
+        if ext == '.npz':
+            model_path = os.path.join(model_dir, filename)
+            print(model_path)
+            channels = 3 if 'rgb' in filename else 1
+            model = model_class(channels)
+            size = 64 + model.offset
+            data = np.zeros((1, channels, size, size), dtype=np.float32)
+            x = chainer.Variable(data)
+            chainer.serializers.load_npz(model_path, model)
+
+            params = {}
+            for path, param in model.namedparams():
+                params[path] = param.array
+
+            net = caffe.Net('upresnet10_3.prototxt', caffe.TEST)
+            for key in net.params:
+                l = len(net.params[key])
+                net.params[key][0].data[...] = params[key + '/W']
+                if l >= 2:
+                    net.params[key][1].data[...] = params[key + '/b']
+ 
+            prototxt_path = '{}.prototxt'.format(fname_convert_table[basename])
+            caffemodel_path = '{}.json.caffemodel'.format(fname_convert_table[basename])
+            net.save(caffemodel_path)
+            shutil.copy('upresnet10_3.prototxt', prototxt_path)
+
+if __name__ == '__main__':
+    caffe.init_log(3)
+    main()