def main_train():
"""
训练模型
:return:
"""
print('[INFO] 解析配置...')
config = None
try:
config = process_config('configs/segmention_config.json')
except Exception as e:
print('[Exception] 配置无效, %s' % e)
exit(0)
# np.random.seed(47) # 固定随机数
print('[INFO] 加载数据...')
dataloader = DataLoader(config=config)
dataloader.prepare_dataset()
train_imgs,train_gt=dataloader.get_train_data()
val_imgs,val_gt=dataloader.get_val_data()
print('[INFO] 构造网络...')
model = SegmentionModel(config=config)
#
print('[INFO] 训练网络...')
trainer = SegmentionTrainer(
model=model.model,
data=[train_imgs,train_gt,val_imgs,val_gt],
config=config)
trainer.train()
print('[INFO] 训练完成...')
def main_train():
"""
Training model
:return:
"""
print('[INFO] Reading Configs...')
config = None
try:
config = process_config('configs/segmention_config.json')
except Exception as e:
print('[Exception] Config Error, %s' % e)
exit(0)
# np.random.seed(47) # 固定随机数
print('[INFO] Preparing Data...')
dataloader = DataLoader(config=config)
dataloader.prepare_dataset()
train_imgs, train_gt = dataloader.get_train_data()
val_imgs, val_gt = dataloader.get_val_data()
print('[INFO] Building Model...')
model = SegmentionModel(config=config)
#
print('[INFO] Training...')
trainer = SegmentionTrainer(
model=model.model,
data=[train_imgs, train_gt, val_imgs, val_gt],
config=config)
trainer.train()
print('[INFO] Finishing...')
def main_test():
print('[INFO] Reading Config file...')
config = None
try:
config = process_config(
'/home/dgxuser102/team34/experiments/configs/segmention_config.json'
)
except Exception as e:
print('[Exception] Config Error, %s' % e)
exit(0)
if repredict == True:
print('[INFO] Making predictions...')
infer = SegmentionInfer(config)
infer.predict()
print('[INFO] Metric results...')
gtlist = fileList(config.test_gt_path, '*' + config.test_gt_datatype)
problist = fileList(config.test_result_path, '*.bmp')
modelName = ['DenseNet-Unet']
drawCurve(gtlist, [problist], modelName, 'DRIVE', config.checkpoint)
print('[INFO] Finished....')
def main_train():
print('[INFO] Reading configuration files')
config = None
try:
config = process_config(
'/home/dgxuser102/data/team34/experiments/configs/segmention_config.json'
)
except Exception as e:
print('[Exception] Configuration Error, %s' % e)
exit(0)
# np.random.seed(47)
print('[INFO] Preparing Data...')
dataloader = DataLoader(config=config)
dataloader.prepare_dataset()
train_imgs, train_gt = dataloader.get_train_data()
val_imgs, val_gt = dataloader.get_val_data()
print('[INFO] Using our model to train...')
model = SegmentionModel(config=config)
#
print('[INFO] Now Training...')
trainer = SegmentionTrainer(model=model.model,
data=[train_imgs, train_gt, val_imgs, val_gt],
config=config)
trainer.train()
print('[INFO] Finishing the training...')
def main_test():
print('[INFO] 解析配置...')
config = None
try:
config = process_config('configs/segmention_config.json')
except Exception as e:
print('[Exception] 配置无效, %s' % e)
exit(0)
print('[INFO] 预测数据...')
infer = SegmentionInfer(config)
infer.predict()
print('[INFO] 预测完成...')
def main_test():
print('[INFO] Reading Configs...')
config = None
try:
config = process_config('configs/segmention_config.json')
except Exception as e:
print('[Exception] Config Error, %s' % e)
exit(0)
if repredict==True:
print('[INFO] Predicting...')
infer = SegmentionInfer( config)
infer.predict()
print('[INFO] Metric results...')
gtlist=fileList(config.test_gt_path,'*'+config.test_gt_datatype)
problist=fileList(config.test_result_path,'*.bmp')
modelName=['Attention-Unet']
drawCurve(gtlist,[problist],modelName,'DRIVE',config.checkpoint)
print('[INFO] Fininshed...')
# -*- coding: utf-8 -*-
"""
Created on Sun Aug 18 11:48:13 2019
@author: kasy
"""
import glob
import cv2
import numpy as np
from tqdm import tqdm
import matplotlib.pyplot as plt
from configs.utils.config_utils import process_config
import os
config = process_config('configs/segmention_config.json')
DATASET_NAME = config.dataset
PRE_NAME = config.preprocess
MODEL_NAME = config.exp_name
#DATASET_NAME = 'DRIVE'
#PRE_NAME = 'DUNet'
#MODEL_NAME = 'VesselNet'
print('Process ', PRE_NAME, ';Dataset ', DATASET_NAME)
#train_dir = './data/{0}/{1}/train/'.format(PRE_NAME, DATASET_NAME)
test_gt_path = './experiments/{0}/test/groundtruth/'.format(MODEL_NAME)
test_prob_path = './experiments/{0}/test/result/{0}/{1}/{2}/result/'.format(MODEL_NAME, PRE_NAME, DATASET_NAME)
def main_test():
print('[INFO] Reading Configs...')
config = None
try:
config = process_config('configs/segmention_config.json')
except Exception as e:
print('[Exception] Config Error, %s' % e)
exit(0)
if repredict==True:
print('[INFO] Predicting...')
infer = SegmentionInfer( config)
infer.predict()
print('[INFO] Metric results...')
# gtlist=fileList(config.test_gt_path,'*'+config.test_gt_datatype)
# problist=fileList(config.test_result_path,'*.png')
# modelName=['DenseNet-Unet']
#drawCurve(gtlist,[problist],modelName,'DRIVE',config.checkpoint)
print('[INFO] Fininshed...')
DATASET_NAME = config.dataset
PRE_NAME = config.preprocess
MODEL_NAME = config.exp_name
# DATASET_NAME = 'DRIVE'
# PRE_NAME = 'DUNet'
# MODEL_NAME = 'VesselNet'
print('Process ', PRE_NAME, ';Dataset ', DATASET_NAME)
# train_dir = './data/{0}/{1}/train/'.format(PRE_NAME, DATASET_NAME)
test_gt_path = './experiments/{0}/test/groundtruth/'.format(MODEL_NAME)
test_prob_path = './experiments/{0}/test/result/{0}/{1}/{2}/result/'.format(MODEL_NAME, PRE_NAME, DATASET_NAME)
# 0_255, gray image
test_gt_list = []
test_prob_list = []
for i in os.listdir(test_gt_path):
test_gt_list.append(os.path.join(test_gt_path, i))
for i in os.listdir(test_prob_path):
# print('---',i)
test_prob_list.append(os.path.join(test_prob_path, i))
# print(test_gt_list)
# print('**', test_prob_list)
# assert 0
gt_img_list = []
prob_img_list = []
for index in range(len(test_gt_list)):
gt_img = cv2.imread(test_gt_list[index])[..., 0]
prob_img = cv2.imread(test_prob_list[index])[..., 0]
gt_img_list.append(gt_img)
prob_img_list.append(prob_img)
TP = []
FP = []
TN = []
FN = []
for threshold in tqdm(range(0, 255)):
temp_TP = 0.0
temp_FP = 0.0
temp_TN = 0.0
temp_FN = 0.0
for index in range(len(test_gt_list)):
prob_img = prob_img_list[index]
gt_img = gt_img_list[index]
# print(prob_img.shape)
gt_img = (gt_img > 0) * 1
prob_img = (prob_img >= threshold) * 1
temp_TP = temp_TP + (np.sum(prob_img * gt_img))
temp_FP = temp_FP + np.sum(prob_img * ((1 - gt_img)))
temp_FN = temp_FN + np.sum(((1 - prob_img)) * ((gt_img)))
temp_TN = temp_TN + np.sum(((1 - prob_img)) * (1 - gt_img))
TP.append(temp_TP)
FP.append(temp_FP)
TN.append(temp_TN)
FN.append(temp_FN)
TP = np.asarray(TP).astype('float32')
FP = np.asarray(FP).astype('float32')
FN = np.asarray(FN).astype('float32')
TN = np.asarray(TN).astype('float32')
FPR = (FP) / (FP + TN)
TPR = (TP) / (TP + FN)
AUC = np.round(np.sum((TPR[1:] + TPR[:-1]) * (FPR[:-1] - FPR[1:])) / 2., 4)
Precision = (TP) / (TP + FP)
Recall = TP / (TP + FN)
MAP = np.round(np.sum((Precision[1:] + Precision[:-1]) * (Recall[:-1] - Recall[1:])) / 2., 4)
# IOU
intersection = 0.0
union = 0.0
threshold = 128
accuracy_list = []
for index in range(len(test_gt_list)):
gt_img = gt_img_list[index]
prob_img = prob_img_list[index]
gt_img = (gt_img > 0) * 1
prob_img = (prob_img >= threshold) * 1
w, h = prob_img.shape[0], prob_img.shape[1]
accuracy = (np.sum(prob_img * gt_img) + np.sum((1 - prob_img) * (1 - gt_img))) / (w * h)
accuracy_list.append(accuracy)
intersection = intersection + np.sum(gt_img * prob_img)
union = union + np.sum(gt_img) + np.sum(prob_img) - np.sum(gt_img * prob_img)
iou = np.round(intersection / union, 4)
pre_median = Precision[128]
recal_median = Recall[128]
beta_square = 0.3
F_beta = ((1 + beta_square) * pre_median * recal_median) / (beta_square * pre_median + recal_median)
print('AUC {0:.4f}, Acc {2:.4f}, F-beta {1:.4f}'.format(AUC, F_beta, np.mean(accuracy_list)))