fix colab link error

LICENSE

@@ -0,0 +1,29 @@
+BSD 3-Clause License
+
+Copyright (c) 2021, Xintao Wang
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its
+   contributors may be used to endorse or promote products derived from
+   this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

README.md

@@ -5,13 +5,16 @@
 [![LICENSE](https://img.shields.io/github/license/xinntao/Real-ESRGAN.svg)](https://github.com/xinntao/Real-ESRGAN/blob/master/LICENSE)
 [![python lint](https://github.com/xinntao/Real-ESRGAN/actions/workflows/pylint.yml/badge.svg)](https://github.com/xinntao/Real-ESRGAN/blob/master/.github/workflows/pylint.yml)
 
-1. [Colab Demo](https://colab.research.google.com/drive/1sVsoBd9AjckIXThgtZhGrHRfFI6UUYOo) for Real-ESRGAN <a href="https://colab.research.google.com/drive/1k2Zod6kSHEvraybHl50Lys0LerhyTMCo?usp=sharing"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="google colab logo"></a>.
+1. [Colab Demo](https://colab.research.google.com/drive/1k2Zod6kSHEvraybHl50Lys0LerhyTMCo?usp=sharing) for Real-ESRGAN <a href="https://colab.research.google.com/drive/1k2Zod6kSHEvraybHl50Lys0LerhyTMCo?usp=sharing"><img src="https://colab.research.google.com/assets/colab-badge.svg" alt="google colab logo"></a>.
 2. [Portable Windows executable file](https://github.com/xinntao/Real-ESRGAN/releases). You can find more information [here](#Portable-executable-files).
 
 Real-ESRGAN aims at developing **Practical Algorithms for General Image Restoration**.<br>
 We extend the powerful ESRGAN to a practical restoration application (namely, Real-ESRGAN), which is trained with pure synthetic data.
 
-:triangular_flag_on_post: The training codes have been released. A detailed guide can be found in [Training.md](Training.md).
+:triangular_flag_on_post: **Updates**
+
+- :white_check_mark: [The inference code](inference_realesrgan.py) supports: 1) **tile** options; 2) images with **alpha channel**; 3) **gray** images; 4) **16-bit** images.
+- :white_check_mark: The training codes have been released. A detailed guide can be found in [Training.md](Training.md).
 
 ### :book: Real-ESRGAN: Training Real-World Blind Super-Resolution with Pure Synthetic Data
 
@@ -49,7 +52,7 @@ If you have some images that Real-ESRGAN could not well restored, please also op
 
 ### Portable executable files
 
-You can download **Windows executable files** from https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN-ncnn-vulkan.zip
+You can download **Windows executable files** from https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.1/RealESRGAN-ncnn-vulkan-20210725-windows.zip
 
 This executable file is **portable** and includes all the binaries and models required. No CUDA or PyTorch environment is needed.<br>
 
@@ -59,6 +62,14 @@ You can simply run the following command:
 ./realesrgan-ncnn-vulkan.exe -i input.jpg -o output.png
 ```
 
+We have provided three models:
+
+1. realesrgan-x4plus  (default)
+2. realesrnet-x4plus
+3. esrgan-x4
+
+You can use the `-n` argument for other models, for example, `./realesrgan-ncnn-vulkan.exe -i input.jpg -o output.png -n realesrnet-x4plus`
+
 Note that it may introduce block inconsistency (and also generate slightly different results from the PyTorch implementation), because this executable file first crops the input image into several tiles, and then processes them separately, finally stitches together.
 
 This executable file is based on the wonderful [Tencent/ncnn](https://github.com/Tencent/ncnn) and [realsr-ncnn-vulkan](https://github.com/nihui/realsr-ncnn-vulkan) by [nihui](https://github.com/nihui).
@@ -106,6 +117,12 @@ python inference_realesrgan.py --model_path experiments/pretrained_models/RealES
 
 Results are in the `results` folder
 
+## :european_castle: Model Zoo
+
+- [RealESRGAN-x4plus](https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x4plus.pth)
+- [RealESRNet-x4plus](https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.1/RealESRNet_x4plus.pth)
+- [official ESRGAN-x4](https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.1/ESRGAN_SRx4_DF2KOST_official-ff704c30.pth)
+
 ## :computer: Training
 
 A detailed guide can be found in [Training.md](Training.md).

Training.md

@@ -34,7 +34,10 @@ DF2K_HR_sub/000001_s003.png
 
 ## Train Real-ESRNet
 
-1. Download pre-trained model [ESRGAN](https://drive.google.com/file/d/1b3_bWZTjNO3iL2js1yWkJfjZykcQgvzT/view?usp=sharing) into `experiments/pretrained_models`.
+1. Download pre-trained model [ESRGAN](https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.1/ESRGAN_SRx4_DF2KOST_official-ff704c30.pth) into `experiments/pretrained_models`.
+    ```bash
+    wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.1/ESRGAN_SRx4_DF2KOST_official-ff704c30.pth -P experiments/pretrained_models
+    ```
 1. Modify the content in the option file `options/train_realesrnet_x4plus.yml` accordingly:
     ```yml
     train:

inference_realesrgan.py

@@ -1,6 +1,7 @@
 import argparse
 import cv2
 import glob
+import math
 import numpy as np
 import os
 import torch
@@ -10,59 +11,233 @@ from torch.nn import functional as F
 
 def main():
     parser = argparse.ArgumentParser()
-    parser.add_argument('--model_path', type=str, default='experiments/pretrained_models/RealESRGAN_x4plus.pth')
+    parser.add_argument('--input', type=str, default='inputs', help='Input image or folder')
-    parser.add_argument('--scale', type=int, default=4)
+    parser.add_argument(
-    parser.add_argument('--input', type=str, default='inputs', help='input image or folder')
+        '--model_path',
+        type=str,
+        default='experiments/pretrained_models/RealESRGAN_x4plus.pth',
+        help='Path to the pre-trained model')
+    parser.add_argument('--scale', type=int, default=4, help='Upsample scale factor')
+    parser.add_argument('--suffix', type=str, default='out', help='Suffix of the restored image')
+    parser.add_argument('--tile', type=int, default=0, help='Tile size, 0 for no tile during testing')
+    parser.add_argument('--tile_pad', type=int, default=10, help='Tile padding')
+    parser.add_argument('--pre_pad', type=int, default=0, help='Pre padding size at each border')
+    parser.add_argument(
+        '--alpha_upsampler',
+        type=str,
+        default='realesrgan',
+        help='The upsampler for the alpha channels. Options: realesrgan | bicubic')
+    parser.add_argument(
+        '--ext',
+        type=str,
+        default='auto',
+        help='Image extension. Options: auto | jpg | png, auto means using the same extension as inputs')
     args = parser.parse_args()
 
-    device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+    upsampler = RealESRGANer(
-    # set up model
+        scale=args.scale, model_path=args.model_path, tile=args.tile, tile_pad=args.tile_pad, pre_pad=args.pre_pad)
-    model = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32, scale=args.scale)
-    loadnet = torch.load(args.model_path)
-    model.load_state_dict(loadnet['params_ema'], strict=True)
-    model.eval()
-    model = model.to(device)
-
     os.makedirs('results/', exist_ok=True)
-    for idx, path in enumerate(sorted(glob.glob(os.path.join(args.input, '*')))):
+    if os.path.isfile(args.input):
-        imgname = os.path.splitext(os.path.basename(path))[0]
+        paths = [args.input]
+    else:
+        paths = sorted(glob.glob(os.path.join(args.input, '*')))
+
+    for idx, path in enumerate(paths):
+        imgname, extension = os.path.splitext(os.path.basename(path))
         print('Testing', idx, imgname)
-        # read image
-        img = cv2.imread(path, cv2.IMREAD_COLOR).astype(np.float32) / 255.
-        img = torch.from_numpy(np.transpose(img[:, :, [2, 1, 0]], (2, 0, 1))).float()
-        img = img.unsqueeze(0).to(device)
 
-        if args.scale == 2:
+        # ------------------------------ read image ------------------------------ #
-            mod_scale = 2
+        img = cv2.imread(path, cv2.IMREAD_UNCHANGED).astype(np.float32)
-        elif args.scale == 1:
+        if np.max(img) > 255:  # 16-bit image
-            mod_scale = 4
+            max_range = 65535
+            print('\tInput is a 16-bit image')
         else:
-            mod_scale = None
+            max_range = 255
-        if mod_scale is not None:
+        img = img / max_range
-            h_pad, w_pad = 0, 0
+        if len(img.shape) == 2:  # gray image
-            _, _, h, w = img.size()
+            img_mode = 'L'
-            if (h % mod_scale != 0):
+            img = cv2.cvtColor(img, cv2.COLOR_GRAY2RGB)
-                h_pad = (mod_scale - h % mod_scale)
+        elif img.shape[2] == 4:  # RGBA image with alpha channel
-            if (w % mod_scale != 0):
+            img_mode = 'RGBA'
-                w_pad = (mod_scale - w % mod_scale)
+            alpha = img[:, :, 3]
-            img = F.pad(img, (0, w_pad, 0, h_pad), 'reflect')
+            img = img[:, :, 0:3]
+            img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
+            if args.alpha_upsampler == 'realesrgan':
+                alpha = cv2.cvtColor(alpha, cv2.COLOR_GRAY2RGB)
+        else:
+            img_mode = 'RGB'
+            img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
 
+        # ------------------- process image (without the alpha channel) ------------------- #
+        upsampler.pre_process(img)
+        if args.tile:
+            upsampler.tile_process()
+        else:
+            upsampler.process()
+        output_img = upsampler.post_process()
+        output_img = output_img.data.squeeze().float().cpu().clamp_(0, 1).numpy()
+        output_img = np.transpose(output_img[[2, 1, 0], :, :], (1, 2, 0))
+        if img_mode == 'L':
+            output_img = cv2.cvtColor(output_img, cv2.COLOR_BGR2GRAY)
+
+        # ------------------- process the alpha channel if necessary ------------------- #
+        if img_mode == 'RGBA':
+            if args.alpha_upsampler == 'realesrgan':
+                upsampler.pre_process(alpha)
+                if args.tile:
+                    upsampler.tile_process()
+                else:
+                    upsampler.process()
+                output_alpha = upsampler.post_process()
+                output_alpha = output_alpha.data.squeeze().float().cpu().clamp_(0, 1).numpy()
+                output_alpha = np.transpose(output_alpha[[2, 1, 0], :, :], (1, 2, 0))
+                output_alpha = cv2.cvtColor(output_alpha, cv2.COLOR_BGR2GRAY)
+            else:
+                h, w = alpha.shape[0:2]
+                output_alpha = cv2.resize(alpha, (w * args.scale, h * args.scale), interpolation=cv2.INTER_LINEAR)
+
+            # merge the alpha channel
+            output_img = cv2.cvtColor(output_img, cv2.COLOR_BGR2BGRA)
+            output_img[:, :, 3] = output_alpha
+
+        # ------------------------------ save image ------------------------------ #
+        if args.ext == 'auto':
+            extension = extension[1:]
+        else:
+            extension = args.ext
+        if img_mode == 'RGBA':  # RGBA images should be saved in png format
+            extension = 'png'
+        save_path = f'results/{imgname}_{args.suffix}.{extension}'
+        if max_range == 65535:  # 16-bit image
+            output = (output_img * 65535.0).round().astype(np.uint16)
+        else:
+            output = (output_img * 255.0).round().astype(np.uint8)
+        cv2.imwrite(save_path, output)
+
+
+class RealESRGANer():
+
+    def __init__(self, scale, model_path, tile=0, tile_pad=10, pre_pad=10):
+        self.scale = scale
+        self.tile_size = tile
+        self.tile_pad = tile_pad
+        self.pre_pad = pre_pad
+        self.mod_scale = None
+
+        # initialize model
+        self.device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
+        model = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32, scale=scale)
+        loadnet = torch.load(model_path)
+        if 'params_ema' in loadnet:
+            keyname = 'params_ema'
+        else:
+            keyname = 'params'
+        model.load_state_dict(loadnet[keyname], strict=True)
+        model.eval()
+        self.model = model.to(self.device)
+
+    def pre_process(self, img):
+        img = torch.from_numpy(np.transpose(img, (2, 0, 1))).float()
+        self.img = img.unsqueeze(0).to(self.device)
+
+        # pre_pad
+        if self.pre_pad != 0:
+            self.img = F.pad(self.img, (0, self.pre_pad, 0, self.pre_pad), 'reflect')
+        # mod pad
+        if self.scale == 2:
+            self.mod_scale = 2
+        elif self.scale == 1:
+            self.mod_scale = 4
+        if self.mod_scale is not None:
+            self.mod_pad_h, self.mod_pad_w = 0, 0
+            _, _, h, w = self.img.size()
+            if (h % self.mod_scale != 0):
+                self.mod_pad_h = (self.mod_scale - h % self.mod_scale)
+            if (w % self.mod_scale != 0):
+                self.mod_pad_w = (self.mod_scale - w % self.mod_scale)
+            self.img = F.pad(self.img, (0, self.mod_pad_w, 0, self.mod_pad_h), 'reflect')
+
+    def process(self):
         try:
             # inference
             with torch.no_grad():
-                output = model(img)
+                self.output = self.model(self.img)
-            # remove extra pad
-            if mod_scale is not None:
-                _, _, h, w = output.size()
-                output = output[:, :, 0:h - h_pad, 0:w - w_pad]
-            # save image
-            output = output.data.squeeze().float().cpu().clamp_(0, 1).numpy()
-            output = np.transpose(output[[2, 1, 0], :, :], (1, 2, 0))
-            output = (output * 255.0).round().astype(np.uint8)
-            cv2.imwrite(f'results/{imgname}_RealESRGAN.png', output)
         except Exception as error:
             print('Error', error)
 
+    def tile_process(self):
+        """Modified from: https://github.com/ata4/esrgan-launcher
+        """
+        batch, channel, height, width = self.img.shape
+        output_height = height * self.scale
+        output_width = width * self.scale
+        output_shape = (batch, channel, output_height, output_width)
+
+        # start with black image
+        self.output = self.img.new_zeros(output_shape)
+        tiles_x = math.ceil(width / self.tile_size)
+        tiles_y = math.ceil(height / self.tile_size)
+
+        # loop over all tiles
+        for y in range(tiles_y):
+            for x in range(tiles_x):
+                # extract tile from input image
+                ofs_x = x * self.tile_size
+                ofs_y = y * self.tile_size
+                # input tile area on total image
+                input_start_x = ofs_x
+                input_end_x = min(ofs_x + self.tile_size, width)
+                input_start_y = ofs_y
+                input_end_y = min(ofs_y + self.tile_size, height)
+
+                # input tile area on total image with padding
+                input_start_x_pad = max(input_start_x - self.tile_pad, 0)
+                input_end_x_pad = min(input_end_x + self.tile_pad, width)
+                input_start_y_pad = max(input_start_y - self.tile_pad, 0)
+                input_end_y_pad = min(input_end_y + self.tile_pad, height)
+
+                # input tile dimensions
+                input_tile_width = input_end_x - input_start_x
+                input_tile_height = input_end_y - input_start_y
+                tile_idx = y * tiles_x + x + 1
+                input_tile = self.img[:, :, input_start_y_pad:input_end_y_pad, input_start_x_pad:input_end_x_pad]
+
+                # upscale tile
+                try:
+                    with torch.no_grad():
+                        output_tile = self.model(input_tile)
+                except Exception as error:
+                    print('Error', error)
+                print(f'\tTile {tile_idx}/{tiles_x * tiles_y}')
+
+                # output tile area on total image
+                output_start_x = input_start_x * self.scale
+                output_end_x = input_end_x * self.scale
+                output_start_y = input_start_y * self.scale
+                output_end_y = input_end_y * self.scale
+
+                # output tile area without padding
+                output_start_x_tile = (input_start_x - input_start_x_pad) * self.scale
+                output_end_x_tile = output_start_x_tile + input_tile_width * self.scale
+                output_start_y_tile = (input_start_y - input_start_y_pad) * self.scale
+                output_end_y_tile = output_start_y_tile + input_tile_height * self.scale
+
+                # put tile into output image
+                self.output[:, :, output_start_y:output_end_y,
+                            output_start_x:output_end_x] = output_tile[:, :, output_start_y_tile:output_end_y_tile,
+                                                                       output_start_x_tile:output_end_x_tile]
+
+    def post_process(self):
+        # remove extra pad
+        if self.mod_scale is not None:
+            _, _, h, w = self.output.size()
+            self.output = self.output[:, :, 0:h - self.mod_pad_h * self.scale, 0:w - self.mod_pad_w * self.scale]
+        # remove prepad
+        if self.pre_pad != 0:
+            _, _, h, w = self.output.size()
+            self.output = self.output[:, :, 0:h - self.pre_pad * self.scale, 0:w - self.pre_pad * self.scale]
+        return self.output
+
 
 if __name__ == '__main__':
     main()

models/realesrgan_model.py

@@ -18,7 +18,7 @@ class RealESRGANModel(SRGANModel):
     def __init__(self, opt):
         super(RealESRGANModel, self).__init__(opt)
         self.jpeger = DiffJPEG(differentiable=False).cuda()
-        self.usm_shaper = USMSharp().cuda()
+        self.usm_sharpener = USMSharp().cuda()
         self.queue_size = opt['queue_size']
 
     @torch.no_grad()
@@ -58,7 +58,7 @@ class RealESRGANModel(SRGANModel):
         if self.is_train:
             # training data synthesis
             self.gt = data['gt'].to(self.device)
-            self.gt_usm = self.usm_shaper(self.gt)
+            self.gt_usm = self.usm_sharpener(self.gt)
 
             self.kernel1 = data['kernel1'].to(self.device)
             self.kernel2 = data['kernel2'].to(self.device)
@@ -160,6 +160,8 @@ class RealESRGANModel(SRGANModel):
 
             # training pair pool
             self._dequeue_and_enqueue()
+            # sharpen self.gt again, as we have changed the self.gt with self._dequeue_and_enqueue
+            self.gt_usm = self.usm_sharpener(self.gt)
         else:
             self.lq = data['lq'].to(self.device)
             if 'gt' in data:

models/realesrnet_model.py

@@ -17,7 +17,7 @@ class RealESRNetModel(SRModel):
     def __init__(self, opt):
         super(RealESRNetModel, self).__init__(opt)
         self.jpeger = DiffJPEG(differentiable=False).cuda()
-        self.usm_shaper = USMSharp().cuda()
+        self.usm_sharpener = USMSharp().cuda()
         self.queue_size = opt['queue_size']
 
     @torch.no_grad()
@@ -59,7 +59,7 @@ class RealESRNetModel(SRModel):
             self.gt = data['gt'].to(self.device)
             # USM the GT images
             if self.opt['gt_usm'] is True:
-                self.gt = self.usm_shaper(self.gt)
+                self.gt = self.usm_sharpener(self.gt)
 
             self.kernel1 = data['kernel1'].to(self.device)
             self.kernel2 = data['kernel2'].to(self.device)

scripts/pytorch2onnx.py

@@ -0,0 +1,17 @@
+import torch
+import torch.onnx
+from basicsr.archs.rrdbnet_arch import RRDBNet
+
+# An instance of your model
+model = RRDBNet(num_in_ch=3, num_out_ch=3, num_feat=64, num_block=23, num_grow_ch=32)
+model.load_state_dict(torch.load('experiments/pretrained_models/RealESRGAN_x4plus.pth')['params_ema'])
+# set the train mode to false since we will only run the forward pass.
+model.train(False)
+model.cpu().eval()
+
+# An example input you would normally provide to your model's forward() method
+x = torch.rand(1, 3, 64, 64)
+
+# Export the model
+with torch.no_grad():
+    torch_out = torch.onnx._export(model, x, 'realesrgan-x4.onnx', opset_version=11, export_params=True)