update

README.md

@@ -7,7 +7,7 @@ and instance semantic segmentation using discriminative loss function for real t
 The main network architecture is as follows:
 
 `Network Architecture`
-![NetWork_Architecture](/data/source_image/network_architecture.png)
+![NetWork_Architecture](./data/source_image/network_architecture.png)
 
 ## Installation
 This software has only been tested on ubuntu 16.04(x64), python3.5, cuda-9.0, cudnn-7.0 with a GTX-1070 GPU. 
@@ -38,23 +38,23 @@ The results are as follows:
 
 `Test Input Image`
 
-![Test Input](/data/tusimple_test_image/0.jpg)
+![Test Input](./data/tusimple_test_image/0.jpg)
 
 `Test Lane Mask Image`
 
-![Test Lane_Mask](/data/source_image/lanenet_mask_result.png)
+![Test Lane_Mask](./data/source_image/lanenet_mask_result.png)
 
 `Test Lane Binary Segmentation Image`
 
-![Test Lane_Binary_Seg](/data/source_image/lanenet_binary_seg.png)
+![Test Lane_Binary_Seg](./data/source_image/lanenet_binary_seg.png)
 
 `Test Lane Instance Segmentation Image`
 
-![Test Lane_Instance_Seg](/data/source_image/lanenet_instance_seg.png)
+![Test Lane_Instance_Seg](./data/source_image/lanenet_instance_seg.png)
 
 `Test Lane Instance Embedding Image`
 
-![Test Lane_Embedding](/data/source_image/lanenet_embedding.png)
+![Test Lane_Embedding](./data/source_image/lanenet_embedding.png)
 
 If you want to test the model on a whole dataset you may call
 ```
@@ -66,7 +66,7 @@ If you set the save_dir argument the result will be saved in that folder or the
 displayed during the inference process holding on 3 seconds per image. I test the model on the whole tusimple lane 
 detection dataset and make it a video. You may catch a glimpse of it bellow.
 `Tusimple test dataset gif`
-![tusimple_batch_test_gif](/data/source_image/lanenet_batch_test.gif)
+![tusimple_batch_test_gif](./data/source_image/lanenet_batch_test.gif)
 
 ## Train your own model
 #### Data Preparation
@@ -99,17 +99,17 @@ python tools/train_lanenet.py --net vgg --dataset_dir data/training_data_example
 You may monitor the training process using tensorboard tools
 
 During my experiment the `Total loss` drops as follows:  
-![Training loss](/data/source_image/total_loss.png)
+![Training loss](./data/source_image/total_loss.png)
 
 The `Binary Segmentation loss` drops as follows:  
-![Training binary_seg_loss](/data/source_image/binary_seg_loss.png)
+![Training binary_seg_loss](./data/source_image/binary_seg_loss.png)
 
 The `Instance Segmentation loss` drops as follows:  
-![Training instance_seg_loss](/data/source_image/instance_seg_loss.png)
+![Training instance_seg_loss](./data/source_image/instance_seg_loss.png)
 
 ## Experiment
 The accuracy during training process rises as follows: 
-![Training accuracy](/data/source_image/accuracy.png)
+![Training accuracy](./data/source_image/accuracy.png)
 
 Please cite my repo [lanenet-lane-detection](https://github.com/MaybeShewill-CV/lanenet-lane-detection) if you use it.
 

config/global_config.py

@@ -1,12 +1,12 @@
 #!/usr/bin/env python3
 # -*- coding: utf-8 -*-
 # @Time    : 18-1-31 上午11:21
-# @Author  : Luo Yao
+# @Author  : MaybeShewill-CV
 # @Site    : https://github.com/MaybeShewill-CV/lanenet-lane-detection
 # @File    : global_config.py
 # @IDE: PyCharm Community Edition
 """
-设置全局变量
+Set global configuration
 """
 from easydict import EasyDict as edict
 
@@ -19,26 +19,25 @@
 __C.TRAIN = edict()
 
 # Set the shadownet training epochs
-__C.TRAIN.EPOCHS = 200010
+__C.TRAIN.EPOCHS = 80010
 # Set the display step
 __C.TRAIN.DISPLAY_STEP = 1
 # Set the test display step during training process
-__C.TRAIN.TEST_DISPLAY_STEP = 1000
+__C.TRAIN.VAL_DISPLAY_STEP = 1000
 # Set the momentum parameter of the optimizer
 __C.TRAIN.MOMENTUM = 0.9
 # Set the initial learning rate
-__C.TRAIN.LEARNING_RATE = 0.0005
+__C.TRAIN.LEARNING_RATE = 0.001
 # Set the GPU resource used during training process
-__C.TRAIN.GPU_MEMORY_FRACTION = 0.85
+__C.TRAIN.GPU_MEMORY_FRACTION = 0.95
 # Set the GPU allow growth parameter during tensorflow training process
 __C.TRAIN.TF_ALLOW_GROWTH = True
 # Set the shadownet training batch size
-__C.TRAIN.BATCH_SIZE = 8
-
+__C.TRAIN.BATCH_SIZE = 4
 # Set the shadownet validation batch size
-__C.TRAIN.VAL_BATCH_SIZE = 8
+__C.TRAIN.VAL_BATCH_SIZE = 4
 # Set the learning rate decay steps
-__C.TRAIN.LR_DECAY_STEPS = 410000
+__C.TRAIN.LR_DECAY_STEPS = 210000
 # Set the learning rate decay rate
 __C.TRAIN.LR_DECAY_RATE = 0.1
 # Set the class numbers
@@ -47,6 +46,16 @@
 __C.TRAIN.IMG_HEIGHT = 256
 # Set the image width
 __C.TRAIN.IMG_WIDTH = 512
+# Set the embedding features dims
+__C.TRAIN.EMBEDDING_FEATS_DIMS = 4
+# Set the random crop pad size
+__C.TRAIN.CROP_PAD_SIZE = 32
+# Set cpu multi process thread nums
+__C.TRAIN.CPU_MULTI_PROCESS_NUMS = 6
+# Set the train moving average decay
+__C.TRAIN.MOVING_AVERAGE_DECAY = 0.9999
+# Set the GPU nums
+__C.TRAIN.GPU_NUM = 2
 
 # Test options
 __C.TEST = edict()
@@ -56,4 +65,14 @@
 # Set the GPU allow growth parameter during tensorflow testing process
 __C.TEST.TF_ALLOW_GROWTH = True
 # Set the test batch size
-__C.TEST.BATCH_SIZE = 8
+__C.TEST.BATCH_SIZE = 2
+
+# Test options
+__C.POSTPROCESS = edict()
+
+# Set the post process connect components analysis min area threshold
+__C.POSTPROCESS.MIN_AREA_THRESHOLD = 100
+# Set the post process dbscan search radius threshold
+__C.POSTPROCESS.DBSCAN_EPS = 0.35
+# Set the post process dbscan min samples threshold
+__C.POSTPROCESS.DBSCAN_MIN_SAMPLES = 800