release commit

.gitignore

@@ -0,0 +1,118 @@
+data/
+log/
+log_old/
+weights/
+output/
+visual/
+.vscode
+
+# Mac directory meta file
+.DS_Store
+
+# intellij config
+.idea
+
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+.hypothesis/
+.pytest_cache/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# pyenv
+.python-version
+
+# celery beat schedule file
+celerybeat-schedule
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2020 Ho Kei Cheng, Jihoon Chung
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -1,14 +1,12 @@
 # CascadePSP: Toward Class-Agnostic and Very High-Resolution Segmentation via Global and Local Refinement
 
-# [In Construction]
-
 Ho Kei Cheng, Jihoon Chung, Yu-Wing Tai, Chi-Keung Tang
 
-[[Paper]]()
+[[Paper]](http://hkchengad.student.ust.hk/CascadePSP/CascadePSP.pdf)
 
-[[Supplementary Information (Comparisons with DenseCRF included!)]]()
+[[Supplementary Information (Comparisons with DenseCRF included!)]](http://hkchengad.student.ust.hk/CascadePSP/CascadePSP-supp-info.pdf)
 
-[[Supplementary image results]]()
+[[Supplementary image results]](http://hkchengad.student.ust.hk/CascadePSP/CascadePSP-supp-images.pdf)
 
 ## Introduction
 

dataset/__init__.py

@@ -0,0 +1,3 @@
+from .online_dataset import OnlineTransformDataset
+from .offline_dataset import OfflineDataset
+from .split_dataset import SplitTransformDataset

dataset/make_bb_trans.py

@@ -0,0 +1,39 @@
+import numpy as np
+
+def is_bb_overlap(rmin, rmax, cmin, cmax, 
+                    crmin, crmax, ccmin, ccmax):
+
+    is_y_overlap = (rmax > crmin) and (crmax > rmin)
+    is_x_overlap = (cmax > ccmin) and (ccmax > cmin)
+
+    return is_x_overlap and is_y_overlap
+
+def get_bb_position(mask):
+    mask = mask > 0.5
+    rows = np.any(mask, axis=1)
+    cols = np.any(mask, axis=0)
+    rmin, rmax = np.where(rows)[0][[0, -1]]
+    cmin, cmax = np.where(cols)[0][[0, -1]]
+
+    # y_min, y_max, x_min, x_max
+    return rmin, rmax, cmin, cmax
+
+def scale_bb_by(rmin, rmax, cmin, cmax, im_height, im_width, h_scale, w_scale):
+    height = rmax - rmin
+    width = cmax - cmin
+
+    rmin -= h_scale * height / 2
+    rmax += h_scale * height / 2
+    cmin -= w_scale * width / 2
+    cmax += w_scale * width / 2
+
+    rmin = int(max(0, rmin))
+    rmax = int(min(im_height-1, rmax))
+    cmin = int(max(0, cmin))
+    cmax = int(min(im_width-1, cmax))
+
+    # Prevent negative width/height
+    rmax = max(rmin, rmax)
+    cmax = max(cmin, cmax)
+
+    return rmin, rmax, cmin, cmax 

dataset/offline_dataset.py

@@ -0,0 +1,116 @@
+import os
+from os import path
+from torch.utils.data.dataset import Dataset
+from torchvision import transforms, utils
+from torchvision.transforms import functional
+from PIL import Image
+import numpy as np
+import progressbar
+
+from dataset.make_bb_trans import *
+
+class OfflineDataset(Dataset):
+    def __init__(self, root, in_memory=False, need_name=False, resize=False, do_crop=False):
+        self.root = root
+        self.need_name = need_name
+        self.resize = resize
+        self.do_crop = do_crop
+        self.in_memory = in_memory
+
+        imgs = os.listdir(root)
+        imgs = sorted(imgs)
+
+        """
+        There are three kinds of files: _im.png, _seg.png, _gt.png
+        """
+        im_list = [im for im in imgs if 'im' in im[-7:].lower()]
+
+        self.im_list = [path.join(root, im) for im in im_list]
+
+        print('%d images found' % len(self.im_list))
+
+        # Make up some transforms
+        self.im_transform = transforms.Compose([
+            transforms.ToTensor(),
+            transforms.Normalize(
+                mean=[0.485, 0.456, 0.406],
+                std=[0.229, 0.224, 0.225]
+            ),
+        ])
+
+        self.gt_transform = transforms.Compose([
+            transforms.ToTensor(),
+        ])
+
+        self.seg_transform = transforms.Compose([
+            transforms.ToTensor(),
+            transforms.Normalize(
+                mean=[0.5],
+                std=[0.5]
+            ),
+        ])
+
+        if self.resize:
+            self.resize_bi = lambda x: x.resize((224, 224), Image.BILINEAR)
+            self.resize_nr = lambda x: x.resize((224, 224), Image.NEAREST)
+        else:
+            self.resize_bi = lambda x: x
+            self.resize_nr = lambda x: x
+
+        if self.in_memory:
+            print('Loading things into memory')
+            self.images = []
+            self.gts = []
+            self.segs = []
+            for im in progressbar.progressbar(self.im_list):
+                image, seg, gt = self.load_tuple(im)
+
+                self.images.append(image)
+                self.segs.append(seg)
+                self.gts.append(gt)
+
+    def load_tuple(self, im):
+        seg = Image.open(im[:-7]+'_seg.png').convert('L')
+        crop_lambda = self.get_crop_lambda(seg)
+
+        image = self.resize_bi(crop_lambda(Image.open(im).convert('RGB')))
+        gt = self.resize_bi(crop_lambda(Image.open(im[:-7]+'_gt.png').convert('L')))
+        seg = self.resize_bi(crop_lambda(Image.open(im[:-7]+'_seg.png').convert('L')))
+
+        return image, seg, gt
+
+    def get_crop_lambda(self, seg):
+        if self.do_crop:
+            seg = np.array(seg)
+            h, w = seg.shape
+            try:
+                bb = get_bb_position(seg)
+                rmin, rmax, cmin, cmax = scale_bb_by(*bb, h, w, 0.15, 0.15)
+                return lambda x: functional.crop(x, rmin, cmin, rmax-rmin, cmax-cmin)
+            except:
+                return lambda x: x
+        else:
+            return lambda x: x
+
+    def __getitem__(self, idx):
+        if self.in_memory:
+            im = self.images[idx]
+            gt = self.gts[idx]
+            seg = self.segs[idx]
+        else:
+            im, seg, gt = self.load_tuple(self.im_list[idx])
+
+        im = self.im_transform(im)
+        gt = self.gt_transform(gt)
+        seg = self.seg_transform(seg)
+
+        if self.need_name:
+            return im, seg, gt, os.path.basename(self.im_list[idx][:-7])
+        else:
+            return im, seg, gt
+
+    def __len__(self):
+        return len(self.im_list)
+
+if __name__ == '__main__':
+    o = OfflineDataset('data/val_static')