进入prepare_data
文件夹打开gen_hard_example.py
脚本,该脚本既可用于生成RNet,也可用于ONet,只要把主函数中:net = 'RNet' 改为 net = 'ONet'。该脚本只生成pos,neg,part三类样本。至于landmark样本,由gen_landmark_aug_24.py和gen_landmark_aug_48.py脚本生成
gen_hard_example.py
代码如下:
#coding:utf-8
import sys
#sys.path.append("../")
from prepare_data.utils import convert_to_square
sys.path.insert(0,'..')
import numpy as np
import argparse
import os
import pickle as pickle
import cv2
from train_models.mtcnn_model import P_Net, R_Net, O_Net
from train_models.MTCNN_config import config
from prepare_data.loader import TestLoader
from Detection.detector import Detector
from Detection.fcn_detector import FcnDetector
from Detection.MtcnnDetector import MtcnnDetector
from utils import *
from prepare_data.data_utils import *
#net : 24(RNet)/48(ONet)
#data: dict()
def save_hard_example(net, data,save_path):
# load ground truth from annotation file
# format of each line: image/path [x1,y1,x2,y2] for each gt_box in this image
im_idx_list = data['images']
# print(images[0])
gt_boxes_list = data['bboxes']
num_of_images = len(im_idx_list)
print("processing %d images in total" % num_of_images)
# save files
neg_label_file = "../../DATA/no_LM%d/neg_%d.txt" % (net, image_size)
neg_file = open(neg_label_file, 'w')
pos_label_file = "../../DATA/no_LM%d/pos_%d.txt" % (net, image_size)
pos_file = open(pos_label_file, 'w')
part_label_file = "../../DATA/no_LM%d/part_%d.txt" % (net, image_size)
part_file = open(part_label_file, 'w')
#read detect result
det_boxes = pickle.load(open(os.path.join(save_path, 'detections.pkl'), 'rb'))
# print(len(det_boxes), num_of_images)
print(len(det_boxes))
print(num_of_images)
assert len(det_boxes) == num_of_images, "incorrect detections or ground truths"
# index of neg, pos and part face, used as their image names
n_idx = 0
p_idx = 0
d_idx = 0
image_done = 0
#im_idx_list image index(list)
#det_boxes detect result(list)
#gt_boxes_list gt(list)
for im_idx, dets, gts in zip(im_idx_list, det_boxes, gt_boxes_list):
gts = np.array(gts, dtype=np.float32).reshape(-1, 4)
if image_done % 100 == 0:
print("%d images done" % image_done)
image_done += 1
if dets.shape[0] == 0:
continue
img = cv2.imread(im_idx)
#change to square
dets = convert_to_square(dets)
dets[:, 0:4] = np.round(dets[:, 0:4])
neg_num = 0
for box in dets:
x_left, y_top, x_right, y_bottom, _ = box.astype(int)
width = x_right - x_left + 1
height = y_bottom - y_top + 1
# ignore box that is too small or beyond image border
if width < 20 or x_left < 0 or y_top < 0 or x_right > img.shape[1] - 1 or y_bottom > img.shape[0] - 1:
continue
# compute intersection over union(IoU) between current box and all gt boxes
Iou = IoU(box, gts)
cropped_im = img[y_top:y_bottom + 1, x_left:x_right + 1, :]
resized_im = cv2.resize(cropped_im, (image_size, image_size),
interpolation=cv2.INTER_LINEAR)
# save negative images and write label
# Iou with all gts must below 0.3
if np.max(Iou) < 0.3 and neg_num < 60:
#save the examples
save_file = get_path(neg_dir, "%s.jpg" % n_idx)
# print(save_file)
neg_file.write(save_file + ' 0\n')
cv2.imwrite(save_file, resized_im)
n_idx += 1
neg_num += 1
else:
# find gt_box with the highest iou
idx = np.argmax(Iou)
assigned_gt = gts[idx]
x1, y1, x2, y2 = assigned_gt
# compute bbox reg label
offset_x1 = (x1 - x_left) / float(width)
offset_y1 = (y1 - y_top) / float(height)
offset_x2 = (x2 - x_right) / float(width)
offset_y2 = (y2 - y_bottom) / float(height)
# save positive and part-face images and write labels
if np.max(Iou) >= 0.65:
save_file = get_path(pos_dir, "%s.jpg" % p_idx)
pos_file.write(save_file + ' 1 %.2f %.2f %.2f %.2f\n' % (
offset_x1, offset_y1, offset_x2, offset_y2))
cv2.imwrite(save_file, resized_im)
p_idx += 1
elif np.max(Iou) >= 0.4:
save_file = os.path.join(part_dir, "%s.jpg" % d_idx)
part_file.write(save_file + ' -1 %.2f %.2f %.2f %.2f\n' % (
offset_x1, offset_y1, offset_x2, offset_y2))
cv2.imwrite(save_file, resized_im)
d_idx += 1
neg_file.close()
part_file.close()
pos_file.close()
def t_net(prefix, epoch,
batch_size, test_mode="PNet",
thresh=[0.6, 0.6, 0.7], min_face_size=25,
stride=2, slide_window=False, shuffle=False, vis=False):
# prefix:['../data/MTCNN_model/PNet_Landmark/PNet', '../data/MTCNN_model/RNet_Landmark/RNet', '../data/MTCNN_model/ONet_Landmark/ONet']
# epoch:[18, 14, 16]
# batch_size:[2048, 256, 16]
# test_mode:"PNet"
# thresh:[0.3, 0.1, 0.7]
# min_face_size:20
# stride=2
# slide_window:False
# shuffle:False
# vis:False
detectors = [None, None, None]
print("Test model: ", test_mode)
#PNet-echo
model_path = ['%s-%s' % (x, y) for x, y in zip(prefix, epoch)] #训练得到的模型文件prefix
print(model_path[0])
# load pnet model
if slide_window:
PNet = Detector(P_Net, 12, batch_size[0], model_path[0]) #2048
else:
PNet = FcnDetector(P_Net, model_path[0])
detectors[0] = PNet
# load rnet model
if test_mode in ["RNet", "ONet"]:
print("==================================", test_mode)
RNet = Detector(R_Net, 24, batch_size[1], model_path[1]) #256
detectors[1] = RNet
# load onet model
if test_mode == "ONet":
print("==================================", test_mode)
ONet = Detector(O_Net, 48, batch_size[2], model_path[2]) #16
detectors[2] = ONet
basedir = '../../DATA/'
#anno_file
filename = './wider_face_train_bbx_gt.txt'
#read anotation(type:dict), include 'images' and 'bboxes'
data = read_annotation(basedir,filename) #read_annotation在data_utils.py中,返回字典data,包含图像路径,以及图像中的人脸框
mtcnn_detector = MtcnnDetector(detectors=detectors, min_face_size=min_face_size,
stride=stride, threshold=thresh, slide_window=slide_window)
print("==================================")
# 注意是在“test”模式下
# imdb = IMDB("wider", image_set, root_path, dataset_path, 'test')
# gt_imdb = imdb.gt_imdb()
print('load test data')
test_data = TestLoader(data['images']) #loader.py,
print ('finish loading')
#list
print ('start detecting....')
detections,_ = mtcnn_detector.detect_face(test_data)
print ('finish detecting ')
save_net = 'RNet'
if test_mode == "PNet":
save_net = "RNet"
elif test_mode == "RNet":
save_net = "ONet"
#save detect result
save_path = os.path.join(data_dir, save_net)
print ('save_path is :')
print(save_path)
if not os.path.exists(save_path):
os.mkdir(save_path)
save_file = os.path.join(save_path, "detections.pkl")
with open(save_file, 'wb') as f:
pickle.dump(detections, f,1)
print("%s测试完成开始OHEM" % image_size)
save_hard_example(image_size, data, save_path)
#命令解析器,定义了一系列的参数,每个参数里面的'help'是该参数的具体描述
def parse_args():
parser = argparse.ArgumentParser(description='Test mtcnn',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('--test_mode', dest='test_mode', help='test net type, can be pnet, rnet or onet',
default='RNet', type=str)
parser.add_argument('--prefix', dest='prefix', help='prefix of model name', nargs="+",
default=['../data/MTCNN_model/PNet_No_Landmark/PNet', '../data/MTCNN_model/RNet_No_Landmark/RNet', '../data/MTCNN_model/ONet_No_Landmark/ONet'],
type=str)
parser.add_argument('--epoch', dest='epoch', help='epoch number of model to load', nargs="+",
default=[18, 14, 16], type=int)
parser.add_argument('--batch_size', dest='batch_size', help='list of batch size used in prediction', nargs="+",
default=[2048, 256, 16], type=int)
parser.add_argument('--thresh', dest='thresh', help='list of thresh for pnet, rnet, onet', nargs="+",
default=[0.3, 0.1, 0.7], type=float)
parser.add_argument('--min_face', dest='min_face', help='minimum face size for detection',
default=20, type=int)
parser.add_argument('--stride', dest='stride', help='stride of sliding window',
default=2, type=int)
parser.add_argument('--sw', dest='slide_window', help='use sliding window in pnet', action='store_true')#只要运行时该变量有传参就将该变量设为True。
#parser.add_argument('--gpu', dest='gpu_id', help='GPU device to train with',default=0, type=int)
parser.add_argument('--shuffle', dest='shuffle', help='shuffle data on visualization', action='store_true')
parser.add_argument('--vis', dest='vis', help='turn on visualization', action='store_true')
args = parser.parse_args()
return args
if __name__ == '__main__':
net = 'ONet'
#net = 'RNet'
if net == "RNet":
image_size = 24
if net == "ONet":
image_size = 48
base_dir = '../../DATA/WIDER_train'
data_dir = '../../DATA/no_LM%s' % str(image_size) #两个数据集的位置
neg_dir = get_path(data_dir, 'negative')
pos_dir = get_path(data_dir, 'positive')
part_dir = get_path(data_dir, 'part')
#创建路径
for dir_path in [neg_dir, pos_dir, part_dir]:
if not os.path.exists(dir_path):
os.makedirs(dir_path)
args = parse_args()
print('Called with argument:')
print(args)
t_net(args.prefix, #训练得到的模型文件
args.epoch, #循环次数
args.batch_size, #测试时使用batch_size,default=[2048, 256, 16]
args.test_mode,#test which model PNet,RNet,ONet
args.thresh, ##分类阈值
args.min_face, #最小脸尺寸为20
args.stride,#滑动框步长
args.slide_window, #在PNet中用滑动窗
args.shuffle,
vis=False)
调用read_annotation函数,read_annotation在data_utils.py中,返回字典data,包含图像路径,以及图像中的人脸框,我们从文件wider_face_train_bbx_gt.txt中读取wider_face数据集的图像,人脸框个数,人脸框位置。它的人脸框保存格式是[x,y,w,h]
注意wider_face_train_bbx_gt.txt和wider_face_train.txt的区别:后者是我们生成PNet样本时用到的文件,它保存人脸框的格式是[x1,y1,x2,y2],如下图所示:
wider_face_train_bbx_gt.txt:
wider_face_train.txt:
def read_annotation(base_dir, label_path): #label_path='./wider_face_train_bbx_gt.txt'
"""
read label file
:param dir: path
:return:
"""
data = dict()
images = []
bboxes = []
labelfile = open(label_path, 'r')
while True:
# image path
imagepath = labelfile.readline().strip('\n') #读取结果:0--Parade/0_Parade_marchingband_1_849.jpg
if not imagepath:
break #跳出循环
imagepath = base_dir + '/WIDER_train/images/' + imagepath
images.append(imagepath)
# face numbers
nums = labelfile.readline().strip('\n') #0--Parade/0_Parade_marchingband_1_849.jpg这个图像的人脸框个数
one_image_bboxes = []
for i in range(int(nums)):
bb_info = labelfile.readline().strip('\n').split(' ')
# only need x, y, w, h
face_box = [float(bb_info[i]) for i in range(4)] #前四个参数就是人脸框的参数
xmin = face_box[0]
ymin = face_box[1]
xmax = xmin + face_box[2]
ymax = ymin + face_box[3]
one_image_bboxes.append([xmin, ymin, xmax, ymax])
bboxes.append(one_image_bboxes)
data['images'] = images#图像路径
data['bboxes'] = bboxes#bboxes是一个列表,每个元素又是一个列表one_image_bboxes,one_image_bboxes列表的元素有一个列表,保存一个矩形框[xmin, ymin, xmax, ymax]
return data
调用了类MtcnnDetector
里面的方法detect_face()
,并传入参数test_data
,代码:
class MtcnnDetector(object):
def __init__(self,
detectors,
min_face_size=20,
stride=2,
threshold=[0.6, 0.7, 0.7],
scale_factor=0.79,
# scale_factor=0.709,#change
slide_window=False):
self.pnet_detector = detectors[0] #初始化,其中一个有,另外两是None
self.rnet_detector = detectors[1]
self.onet_detector = detectors[2]
self.min_face_size = min_face_size
self.stride = stride
self.thresh = threshold
self.scale_factor = scale_factor
self.slide_window = slide_window
def detect_face(self, test_data):
all_boxes = [] # #保存每一张图片的bboxes
landmarks = []
batch_idx = 0
sum_time = 0
t1_sum = 0
t2_sum = 0
t3_sum = 0
num_of_img = test_data.size #图像个数
empty_array = np.array([])
# test_data is iter_
s_time = time.time() #返回当前时间的时间戳
for databatch in test_data:
# databatch(image returned) #依次提取test_data里面的每一张图片,每提取一百张图片,打印进度和所耗费时间
batch_idx += 1
if batch_idx % 100 == 0:
c_time = (time.time() - s_time )/100
print("%d out of %d images done" % (batch_idx ,test_data.size))
print('%f seconds for each image' % c_time)
s_time = time.time()
im = databatch
# pnet
if self.pnet_detector: #如果初始化的参数是PNet
st = time.time()
# ignore landmark
boxes, boxes_c, landmark = self.detect_pnet(im) #这里调用了该类下的方法detect_pnet,
t1 = time.time() - st
sum_time += t1
t1_sum += t1
if boxes_c is None:
print("boxes_c is None...")
all_boxes.append(empty_array)
# pay attention
landmarks.append(empty_array)
continue
#print(all_boxes)
# rnet
if self.rnet_detector:
t = time.time()
# ignore landmark
boxes, boxes_c, landmark = self.detect_rnet(im, boxes_c)# 传入图片和pnet_detector返回的bbox坐标
t2 = time.time() - t
sum_time += t2
t2_sum += t2
if boxes_c is None:
all_boxes.append(empty_array)
landmarks.append(empty_array)
continue
# onet
if self.onet_detector:
t = time.time()
boxes, boxes_c, landmark = self.detect_onet(im, boxes_c)
t3 = time.time() - t
sum_time += t3
t3_sum += t3
if boxes_c is None:
all_boxes.append(empty_array)
landmarks.append(empty_array)
continue
all_boxes.append(boxes_c)
landmark = [1]
landmarks.append(landmark)
print('num of images', num_of_img)
print("time cost in average" +
'{:.3f}'.format(sum_time/num_of_img) +
' pnet {:.3f} rnet {:.3f} onet {:.3f}'.format(t1_sum/num_of_img, t2_sum/num_of_img,t3_sum/num_of_img))
# num_of_data*9,num_of_data*10
print('boxes length:',len(all_boxes))
return all_boxes, landmarks
上面的类MtcnnDetector
里面的方法detect_face()
调用了类MtcnnDetector
里面的方法detect_pnet()
,并传入参数im,:
def detect_pnet(self, im):
"""Get face candidates through pnet
Parameters:
----------
im: numpy array
input image array
Returns:
-------
boxes: numpy array
detected boxes before calibration
boxes_c: numpy array
boxes after calibration
"""
h, w, c = im.shape #获得图片的宽、高、通道数
net_size = 12
current_scale = float(net_size) / self.min_face_size #current_scale=12 / 20 = 0.6
# 我们的卷积核是12*12,假设我们要求检测的最小人脸是min_face_size=6,我们必须要放大至12*12,这样,卷积后才能检测的到,如果直接用12*12卷积,人脸之战6*6,干扰过大,可能识别不出啦
#假设我们要求检测的最小人脸是min_face_size=24,我们必须要缩放至12*12,否则12*12只能覆盖一部分人脸
im_resized = self.processed_image(im, current_scale) #缩放图像0.6倍
current_height, current_width, _ = im_resized.shape
# fcn
all_boxes = list()
while min(current_height, current_width) > net_size:
# return the result predicted by pnet
# cls_cls_map : H*w*2
# reg: H*w*4
# class_prob andd bbox_pred
cls_cls_map, reg = self.pnet_detector.predict(im_resized) #找pnet_detector方法,这个方式用传入的detectors[0]=PNet来初始化,PNet是Detector类封装,所以最终是调用Detector类的predict函数,即self.pnet_detector = detectors[0] = FcnDetector(P_Net, model_path[0])
# 返回PNet网络的预测结果,得到class_prob 和 bbox_pred
# boxes: num*9(x1,y1,x2,y2,score,x1_offset,y1_offset,x2_offset,y2_offset)
boxes = self.generate_bbox(cls_cls_map[:, :, 1], reg, current_scale, self.thresh[0])
#将宽高进一步缩放,形成图像金字塔,注意这里scale_factor默认为0.79,在宽高小于12之前一直进行此while循环
current_scale *= self.scale_factor
im_resized = self.processed_image(im, current_scale) #将im缩放0.79*0.6 倍
current_height, current_width, _ = im_resized.shape #获得新的高宽
if boxes.size == 0:
continue
#从非极大值抑制算法获得index,此处的nms是针对一张图像上的矩形框
keep = py_nms(boxes[:, :5], 0.5, 'Union')
boxes = boxes[keep] #根据索引提取出矩形框
all_boxes.append(boxes)
if len(all_boxes) == 0:
return None, None, None
all_boxes = np.vstack(all_boxes) #按照行顺序把数组给堆叠起来,表示整个金字塔所有图相上检测的矩形框
# 针对金字塔所有矩形框作nms
keep = py_nms(all_boxes[:, 0:5], 0.7, 'Union')
all_boxes = all_boxes[keep]
boxes = all_boxes[:, :5]
bbw = all_boxes[:, 2] - all_boxes[:, 0] + 1
bbh = all_boxes[:, 3] - all_boxes[:, 1] + 1
# 得到bbox的坐标
boxes_c = np.vstack([all_boxes[:, 0] + all_boxes[:, 5] * bbw,
all_boxes[:, 1] + all_boxes[:, 6] * bbh,
all_boxes[:, 2] + all_boxes[:, 7] * bbw,
all_boxes[:, 3] + all_boxes[:, 8] * bbh,
all_boxes[:, 4]])
boxes_c = boxes_c.T
return boxes, boxes_c, None
该函数中调用了 im_resized = self.processed_image(im, current_scale),用于求出图像的初始尺寸
def processed_image(self, img, scale):
'''
rescale/resize the image according to the scale
:param img: image
:param scale:
:return: resized image
'''
height, width, channels = img.shape
new_height = int(height * scale) # resized new height
new_width = int(width * scale) # resized new width
new_dim = (new_width, new_height)
img_resized = cv2.resize(img, new_dim, interpolation=cv2.INTER_LINEAR) # resized image
# don't understand this operation
img_resized = (img_resized - 127.5) / 128
return img_resized
根据最小人脸检测尺寸与size12*12的比值,对原图进行放大或缩小,作为金字塔的底层图像。
detect_pnet()方法
调用了类MtcnnDetector
里面的方法generate_bbox()
,并传入参数cls_cls_map[:, :, 1], reg, current_scale, self.thresh[0]
def generate_bbox(self, cls_map, reg, scale, threshold):
"""
generate bbox from feature cls_map according to the threshold
Parameters:
----------
cls_map: numpy array , n x m
detect score for each position
reg: numpy array , n x m x 4,这里的回归量是偏差量
bbox
scale: float number
scale of this detection
threshold: float number
detect threshold
Returns:
-------
bbox array
"""
stride = 2
# stride = 4
cellsize = 12
# cellsize = 25
#返回人脸分类概率大于0.6的样本的index
t_index = np.where(cls_map > threshold) #[0.6, 0.7, 0.7],这里选第一个PNet=0.6
# #不存在对应的样本时返回空值
if t_index[0].size == 0:
return np.array([])
dx1, dy1, dx2, dy2 = [reg[t_index[0], t_index[1], i] for i in range(4)] # reg= n x m x 4,#得到对应bbox的offset
reg = np.array([dx1, dy1, dx2, dy2])
score = cls_map[t_index[0], t_index[1]] #根据行列点的索引找对应的人脸概率
boundingbox = np.vstack([np.round((stride * t_index[1]) / scale),
np.round((stride * t_index[0]) / scale),
np.round((stride * t_index[1] + cellsize) / scale),
np.round((stride * t_index[0] + cellsize) / scale),
score,
reg])
# 原始图片中回归框坐标需要经过反向运算,计算方式如下,其中cellSize=12,是因为12*12的图片进去后变成1*1
# stride=2是因为几层卷积中只有一个stride为2,scale代表的是我们在哪个尺度金字塔的图像,
return boundingbox.T
detect_pnet()方法还
调用了类FcnDetector
,并传入参数im_resized
class FcnDetector(object):
#net_factory: which net
#model_path: where the params'file is
def __init__(self, net_factory, model_path):
#create a graph
graph = tf.Graph()
with graph.as_default():
##在图中定义张量和运算 graph(-1,1)
self.image_op = tf.placeholder(tf.float32, name='input_image')
self.width_op = tf.placeholder(tf.int32, name='image_width')
self.height_op = tf.placeholder(tf.int32, name='image_height')
image_reshape = tf.reshape(self.image_op, [1, self.height_op, self.width_op, 3])
#self.cls_prob=batch*2
#self.bbox_pred =batch*4
# 调用了PNet网络,得到训练PNet后的self.cls_prob和self.bbox_pred
self.cls_prob, self.bbox_pred, _ = net_factory(image_reshape, training=False)
#allow
self.sess = tf.Session(config=tf.ConfigProto(allow_soft_placement=True, gpu_options=tf.GPUOptions(allow_growth=True)))
saver = tf.train.Saver()
#check whether the dictionary is valid
model_dict = '/'.join(model_path.split('/')[:-1])
ckpt = tf.train.get_checkpoint_state(model_dict)
print(model_path)
readstate = ckpt and ckpt.model_checkpoint_path
assert readstate, "the params dictionary is not valid"
print("restore models' param")
saver.restore(self.sess, model_path)
def predict(self, databatch):
height, width, _ = databatch.shape #输入一张图
# print(height, width)
cls_prob, bbox_pred = self.sess.run([self.cls_prob, self.bbox_pred],
feed_dict={self.image_op: databatch, self.width_op: width,
self.height_op: height})
return cls_prob, bbox_pred
save_hard_example函数用于从PNet输出的检测结果中,提取矩形框保存,并进入TXT信息,作为RNet,ONet的训练样本:
def save_hard_example(net, data,save_path):
# load ground truth from annotation file
# format of each line: image/path [x1,y1,x2,y2] for each gt_box in this image
#net=12
im_idx_list = data['images'] #图像名
gt_boxes_list = data['bboxes'] #人脸框,特征点
num_of_images = len(im_idx_list)
print("processing %d images in total" % num_of_images)
# TXT文件用于写入图像名,label,offset等
neg_label_file = "../../DATA/no_LM%d/neg_%d.txt" % (net, image_size)# ../../DATA/no_LM12/neg_12.txt"
neg_file = open(neg_label_file, 'w')
pos_label_file = "../../DATA/no_LM%d/pos_%d.txt" % (net, image_size)
pos_file = open(pos_label_file, 'w')
part_label_file = "../../DATA/no_LM%d/part_%d.txt" % (net, image_size)
part_file = open(part_label_file, 'w')
#read detect result
det_boxes = pickle.load(open(os.path.join(save_path, 'detections.pkl'), 'rb'))
# print(len(det_boxes), num_of_images)
print(len(det_boxes))
print(num_of_images)
assert len(det_boxes) == num_of_images, "incorrect detections or ground truths"
# index of neg, pos and part face, used as their image names
n_idx = 0
p_idx = 0
d_idx = 0
image_done = 0
for im_idx, dets, gts in zip(im_idx_list, det_boxes, gt_boxes_list): #图像名,检测结果框,真实人脸框
gts = np.array(gts, dtype=np.float32).reshape(-1, 4)
if image_done % 100 == 0:
print("%d images done" % image_done)
image_done += 1
if dets.shape[0] == 0:
continue
img = cv2.imread(im_idx)
#change to square
dets = convert_to_square(dets)
dets[:, 0:4] = np.round(dets[:, 0:4])
neg_num = 0
for box in dets: #dets里保存:是一张图中检测到的多个矩形框,一个一个框的遍历处理
x_left, y_top, x_right, y_bottom, _ = box.astype(int)
width = x_right - x_left + 1
height = y_bottom - y_top + 1
# 忽略过小和超出边界的矩形框
if width < 20 or x_left < 0 or y_top < 0 or x_right > img.shape[1] - 1 or y_bottom > img.shape[0] - 1:
continue
# 计算当前检测的框(1个)与真实框gts(多个)的iou
Iou = IoU(box, gts)
cropped_im = img[y_top:y_bottom + 1, x_left:x_right + 1, :]
resized_im = cv2.resize(cropped_im, (image_size, image_size),
interpolation=cv2.INTER_LINEAR) #缩放成24*24
# 保存负样本,并写入TXT
if np.max(Iou) < 0.3 and neg_num < 60:
#save the examples
save_file = get_path(neg_dir, "%s.jpg" % n_idx) #当前检测的框24*24存入 neg_dir='../../DATA/no_LM24/negative',生成这个路径
# print(save_file)
neg_file.write(save_file + ' 0\n') # 在./../DATA/no_LM12/neg_12.txt"文件中写入:图像路径+label
cv2.imwrite(save_file, resized_im)
n_idx += 1
neg_num += 1
else:
# find gt_box with the highest iou
idx = np.argmax(Iou)
assigned_gt = gts[idx] #真实框
x1, y1, x2, y2 = assigned_gt
# x1是真实值,x_left是检测值,计算偏移量
offset_x1 = (x1 - x_left) / float(width)
offset_y1 = (y1 - y_top) / float(height)
offset_x2 = (x2 - x_right) / float(width)
offset_y2 = (y2 - y_bottom) / float(height)
# 保存正样本和部分样本
if np.max(Iou) >= 0.65:
save_file = get_path(pos_dir, "%s.jpg" % p_idx)
pos_file.write(save_file + ' 1 %.2f %.2f %.2f %.2f\n' % (
offset_x1, offset_y1, offset_x2, offset_y2)) #向TXT文件中写入(路径+label+offset
cv2.imwrite(save_file, resized_im)
p_idx += 1
elif np.max(Iou) >= 0.4: #part样本
save_file = os.path.join(part_dir, "%s.jpg" % d_idx)
part_file.write(save_file + ' -1 %.2f %.2f %.2f %.2f\n' % (
offset_x1, offset_y1, offset_x2, offset_y2))
cv2.imwrite(save_file, resized_im)
d_idx += 1
neg_file.close()
part_file.close()
pos_file.close()
参考:https://blog.csdn.net/He_yuan_hong/article/details/85218240