lanenet

.gitignore

@@ -0,0 +1,88 @@
+# Compiled source #
+###################
+*.com
+*.class
+*.dll
+*.exe
+*.o
+*.so
+*.pyc
+.ipynb_checkpoints
+*~
+*#
+build*
+
+# Packages #
+###################
+# it's better to unpack these files and commit the raw source
+# git has its own built in compression methods
+*.7z
+*.dmg
+*.gz
+*.iso
+*.jar
+*.rar
+*.tar
+*.zip
+
+# Logs and databases #
+######################
+*.log
+*.sql
+*.sqlite
+
+# OS generated files #
+######################
+.DS_Store
+.DS_Store?
+._*
+.Spotlight-V100
+.Trashes
+ehthumbs.db
+Thumbs.db
+
+# Images
+######################
+*.jpg
+*.gif
+*.png
+*.svg
+*.ico
+
+# Video
+######################
+*.wmv
+*.mpg
+*.mpeg
+*.mp4
+*.mov
+*.flv
+*.avi
+*.ogv
+*.ogg
+*.webm
+
+# Audio
+######################
+*.wav
+*.mp3
+*.wma
+
+# Fonts
+######################
+Fonts
+*.eot
+*.ttf
+*.woff
+
+# Format
+######################
+CPPLINT.cfg
+.clang-format
+
+# Gtags
+######################
+GPATH
+GRTAGS
+GSYMS
+GTAGS

README.md

@@ -0,0 +1,15 @@
+## Lanenet with Enet backbone ##
+***
+
+### Dataset ###
+- [TuSimple Lane Detection](https://github.com/TuSimple/tusimple-benchmark/issues/3)
+
+### Test Sample ###
+***
+
+![test sample](./docs/images/result.gif)
+
+### References ###
+- [Towards End-to-End Lane Detection: an Instance Segmentation Approach](https://arxiv.org/pdf/1802.05591.pdf)
+- [Learning Lightweight Lane Detection CNNs by Self Attention Distillation](https://arxiv.org/pdf/1908.00821.pdf)
+- [Semantic Instance Segmentation with a Discriminative Loss Function](https://arxiv.org/pdf/1708.02551.pdf)

config.py

@@ -0,0 +1,29 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+import os
+
+
+_CURRENT_DIR = os.path.dirname(os.path.realpath(__file__))
+
+
+class Config(object):
+    def __init__(self):
+        self.CURRENT_DIR = _CURRENT_DIR
+
+        self.DATA_PATH = os.path.abspath(os.path.join(_CURRENT_DIR, "data"))
+
+        self.SAVED_MODEL_PATH = os.path.abspath(
+            os.path.join(_CURRENT_DIR, "saved_models")
+        )
+        if not os.path.isdir(self.SAVED_MODEL_PATH):
+            os.system("mkdir -p {}".format(self.SAVED_MODEL_PATH))
+
+    def display(self):
+        """
+        Display Configuration values.
+        """
+        print("\nConfigurations:")
+        for a in dir(self):
+            if not a.startswith("__") and not callable(getattr(self, a)):
+                print("{:30} {}".format(a, getattr(self, a)))
+        print("\n")

data_loader/__init__.py

@@ -0,0 +1,4 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+from .tu_simple_dataset import TuSimpleDataset
+from .tu_simple_data_transform import TuSimpleDataTransform

data_loader/data_transform_base.py

@@ -0,0 +1,26 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+from albumentations import Compose
+
+
+class DataTransformBase(object):
+    def __init__(self):
+        self._train_transform_list = []
+        self._val_transform_list = []
+
+        self._transforms_dict = {}
+
+    def _initialize_transform_dict(self):
+        self._transforms_dict["train"] = Compose(self._train_transform_list)
+        self._transforms_dict["val"] = Compose(self._val_transform_list)
+
+    def __call__(self, image, mask=None, phase=None):
+        if phase is None:
+            return self._transforms_dict["val"](image=image)["image"]
+
+        assert phase in ("train", "val")
+        assert mask is not None
+
+        augmented = self._transforms_dict[phase](image=image, masks=[mask])
+
+        return augmented["image"], augmented["masks"][0]

data_loader/tu_simple_data_transform.py

@@ -0,0 +1,45 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+from .data_transform_base import DataTransformBase
+from albumentations import (
+    Resize,
+    HorizontalFlip,
+    Normalize,
+    GaussNoise,
+    RandomBrightnessContrast,
+    RandomShadow,
+    RandomRain,
+    Rotate,
+    ShiftScaleRotate,
+    RandomResizedCrop,
+)
+from albumentations.pytorch import ToTensor
+import random
+
+
+class TuSimpleDataTransform(DataTransformBase):
+    def __init__(self, num_classes, input_size):
+        super(TuSimpleDataTransform, self).__init__()
+
+        random.seed(1000)
+        height, width = input_size
+
+        self._train_transform_list = self._train_transform_list + [
+            HorizontalFlip(p=0.5),
+            GaussNoise(p=0.5),
+            RandomBrightnessContrast(p=0.5),
+            RandomShadow(p=0.5),
+            RandomRain(rain_type="drizzle", p=0.5),
+            ShiftScaleRotate(rotate_limit=10, p=0.5),
+            RandomResizedCrop(
+                height=height, width=width, scale=(0.8, 1), p=0.5
+            ),
+        ]
+
+        self._train_transform_list.append(Resize(height, width))
+        self._val_transform_list.append(Resize(height, width))
+
+        self._train_transform_list.append(ToTensor(num_classes=num_classes))
+        self._val_transform_list.append(ToTensor(num_classes=num_classes))
+
+        self._initialize_transform_dict()