def Generate(self):
n = len(self.labels)
i = 0
clone_labels = self.labels.copy()
while True:
if i==0:
random.shuffle(clone_labels)
# 数据平均
label = clone_labels[i]
# print('image_path:', label['image_path'])
# 读取图片
image_path = label['image_path']
img = image_helper.fileToOpencvImage(image_path)
# print('imgType:', type(img))
# width, height = ImageHelper.opencvGetImageSize(img)
# print('imgSize:', width, height)
# 获取随机变换图片及标签
points = label['points']
img, points = self._get_random_data(
img, target_points=points)
# 缩放图片
img, points, _ = image_helper.opencvProportionalResize(
img, self.input_shape, points=points, bg_color=None, bg_mode=None)
# 最后输出图片
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# 调整参数范围
img =img.astype(np.float32)
img = img / 255
points = points/self.input_shape
points = points[...,::-1]
i = (i+1) % n
if (points<0).any() or (points>1).any():
continue
yield img, points
def get_random_data(self, label):
'''
随机数据扩展
Args:
label: {
'image_path': 图片路径,
'classes': [obj_num,],
'boxes': [obj_num,[x1,y1,x2,y2]]
}
Returns:
random_img: [h,w,3]
random_boxes: [obj_num,[y1,x1,y2,x2]]
classes: [obj_num,]
'''
random_img = ImageHelper.fileToOpencvImage(label['image_path'])
random_boxes = label['boxes']
random_boxes = np.array(random_boxes)
# 转成点集
random_boxes = random_boxes.reshape((-1, 2))
# 图片大小
# width, height = ImageHelpler.opencvGetImageSize(random_img)
# 画矩形
# cv2.rectangle(image, (20, 20), (380, 380), tuple(np.random.randint(0, 30, (3), dtype=np.int32)), thickness=8)
# 增加模糊
ksize = random.randint(0, 4)
if ksize > 0:
random_img = ImageHelper.opencvBlur(random_img, (ksize, ksize))
# 变换图像，旋转会导致框不准
# random_offset_x = random.random()*width-width/2
# random_offset_y = random.random()*height-height/2
random_offset_x = random.random() * 90 - 45
random_offset_y = random.random() * 90 - 45
# random_angle_x = random.random()*60-30
# random_angle_y = random.random()*60-30
# random_angle_z = random.random()*40-20
random_scale_x = random.random() * 1.5 + 0.5
random_scale_y = random.random() * 1.5 + 0.5
random_scale_z = 1
# random_offset_x = 0
# random_offset_y = 0
random_angle_x = 0
random_angle_y = 0
random_angle_z = 0
# random_scale = 1
random_img, org, dst, random_boxes = ImageHelper.opencvPerspective(
random_img,
offset=(random_offset_x, random_offset_y, 0),
angle=(random_angle_x, random_angle_y, random_angle_z),
scale=(random_scale_x, random_scale_y, random_scale_z),
points=random_boxes,
bg_color=None,
bg_mode=None)
# 增加线条
# random_img = image_helper.opencvRandomLines(random_img, 8)
# 增加噪声
random_img = ImageHelper.opencvNoise(random_img)
# 颜色抖动
# random_img = ImageHelper.opencvRandomColor(random_img)
# 调整图片大小
random_img, random_boxes, padding = ImageHelper.opencvProportionalResize(
random_img,
self.image_size,
points=random_boxes,
bg_color=None,
bg_mode=None)
# 最后输出图片
random_img = cv2.cvtColor(random_img, cv2.COLOR_BGR2RGB)
# 调整参数范围
random_img = random_img.astype(np.float32)
random_img = random_img / 255
# 转换boxes成框列表
random_boxes = random_boxes.reshape((-1, 4))
# # 将4角坐标转换成矩形坐标
# random_boxes = np.reshape(random_boxes, (-1, 4, 2))
# boxes_min = np.min(random_boxes, axis=1)
# boxes_max = np.max(random_boxes, axis=1)
# random_boxes = np.concatenate([boxes_min, boxes_max], axis=-1)
# 截取框超出图片部分
random_boxes[:, 0][random_boxes[:, 0] < 0] = 0
random_boxes[:, 1][random_boxes[:, 1] < 0] = 0
random_boxes[:, 2][
random_boxes[:, 2] > self.image_size[0]] = self.image_size[0]
random_boxes[:, 3][
random_boxes[:, 3] > self.image_size[1]] = self.image_size[1]
# 去掉无效框
mask = np.logical_and(random_boxes[:, 2] - random_boxes[:, 0] >= 2,
random_boxes[:, 3] - random_boxes[:, 1] >= 2)
random_boxes = random_boxes[mask][:, [1, 0, 3, 2]]
classes = np.array(label['classes'], dtype=np.int32)
classes = classes[mask]
# cv2.imwrite(path, image)
return random_img, random_boxes, classes
def predict(request):
'''识别'''
request_data = json.loads(request.body)
# print('request_data:', request_data)
read = request_data['read']
img_data = request_data['img_data'].split(',')[1]
img_data = ImageHelper.base64ToBytes(img_data)
img_old = ImageHelper.bytesToOpencvImage(img_data)
# 图片大小
image_size = np.int32([416, 416])
# 缩放图片
img, _, padding = ImageHelper.opencvProportionalResize(img_old,
image_size,
bg_color=(0, 0, 0))
# print('imgType:', type(img))
width, height = ImageHelper.opencvGetImageSize(img_old)
image_size_old = np.int32([width, height])
print('imgSize:', width, height)
# 最后输出图片
predict_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# 调整参数范围
predict_img = predict_img.astype(np.float32)
predict_img = predict_img / 255
# 增加一个维度
predict_img = np.expand_dims(predict_img, 0)
y_boxes, y_classes_id, y_scores, y_classes, y_confidence = model.Predict(
predict_img, confidence_thresh=0.5, scores_thresh=0.2, iou_thresh=0.5)
# 结果转物体框列表
y_boxes = y_boxes.numpy()
y_classes_id = y_classes_id.numpy()
y_scores = y_scores.numpy()
y_classes = y_classes.numpy()
y_confidence = y_confidence.numpy()
y_boxes[:, [0, 2]] = (y_boxes[:, [0, 2]] * image_size[0] - padding[2]) / (
image_size[0] - padding[2] - padding[3]) * image_size_old[0]
y_boxes[:, [1, 3]] = (y_boxes[:, [1, 3]] * image_size[1] - padding[0]) / (
image_size[1] - padding[0] - padding[1]) * image_size_old[1]
# 截取框超出图片部分
y_boxes[:, 0][y_boxes[:, 0] < 0] = 0
y_boxes[:, 1][y_boxes[:, 1] < 0] = 0
y_boxes[:, 2][y_boxes[:, 2] > image_size_old[0]] = image_size_old[0]
y_boxes[:, 3][y_boxes[:, 3] > image_size_old[1]] = image_size_old[1]
# 去掉无效框
y_mask = np.logical_and(y_boxes[:, 2] - y_boxes[:, 0] > 2,
y_boxes[:, 3] - y_boxes[:, 1] > 2)
y_boxes = y_boxes[y_mask]
y_classes_id = y_classes_id[y_mask]
y_scores = y_scores[y_mask]
y_classes = y_classes[y_mask]
y_confidence = y_confidence[y_mask]
y_boxes = y_boxes.astype(np.int32)
print('y_boxes:', y_boxes.shape)
print('y_classes_id:', y_classes_id.shape)
print('y_scores:', y_scores.shape)
print('y_classes:', y_classes.shape)
print('y_confidence:', y_confidence.shape)
# 画框
result_img = img_old.copy()
for i in range(y_boxes.shape[0]):
print('y_boxes:', y_boxes[i, :])
print('y_classes_id:', y_classes_id[i])
print('y_scores:', y_scores[i])
print('y_classes:', y_classes[i])
print('y_confidence:', y_confidence[i])
cv2.rectangle(result_img,
tuple(y_boxes[i, 0:2]),
tuple(y_boxes[i, 2:4]), (0, 0, 255),
thickness=1)
cv2.putText(result_img, classes_name[y_classes_id[i]],
tuple(y_boxes[i, 0:2]), cv2.FONT_HERSHEY_COMPLEX, 0.6,
(0, 100, 0), 1)
cv2.putText(result_img, str(y_scores[i]),
tuple(y_boxes[i, 0:2] + (0, 20)), cv2.FONT_HERSHEY_COMPLEX,
0.6, (0, 0, 100), 1)
jsonObj = {
'boxes':
y_boxes.tolist(),
'classes':
y_classes.tolist(),
'random_img':
ImageHelper.bytesTobase64(ImageHelper.opencvImageToBytes(img)),
'result_img':
ImageHelper.bytesTobase64(ImageHelper.opencvImageToBytes(result_img)),
}
# print('jsonObj:',jsonObj)
return HttpResponse(json.dumps(jsonObj), content_type="application/json")
def test():
'''训练'''
# 加载数据
image_wh = (416, 416)
# image_wh = (608, 608)
anchors = LoadAnchors(anchorsFile)
classes_name, classes_num = LoadClasses(classesFile)
# 构建模型
model = YoloV3Model(classes_num=classes_num,
anchors=anchors,
image_wh=image_wh)
# 编译模型
print('编译模型')
model.compile(optimizer=tf.keras.optimizers.Adam(lr=1e-4))
# 加载模型
_ = model(tf.ones((1, image_wh[1], image_wh[0], 3)), training=False)
if os.path.exists(modelPath):
last_model_path = tf.train.latest_checkpoint(modelPath)
model.load_weights(last_model_path).expect_partial()
print('加载模型:{}'.format(last_model_path))
# model.summary()
img_old = ImageHelper.fileToOpencvImage(imageFile)
# 缩放图片
img, _, padding = ImageHelper.opencvProportionalResize(img_old,
image_wh,
bg_color=(0, 0, 0))
# print('imgType:', type(img))
width, height = ImageHelper.opencvGetImageSize(img_old)
image_size_old = np.int32([width, height])
print('imgSize:', width, height)
# 最后输出图片
predict_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# 调整参数范围
predict_img = predict_img.astype(np.float32)
predict_img = predict_img / 255
# 增加一个维度
predict_img = np.expand_dims(predict_img, 0)
y_boxes, y_classes_id, y_scores, y_classes, y_confidence = model.Predict(
predict_img, confidence_thresh=0.5, scores_thresh=0.2, iou_thresh=0.5)
# 结果转物体框列表
y_boxes = y_boxes.numpy()
y_classes_id = y_classes_id.numpy()
y_scores = y_scores.numpy()
y_classes = y_classes.numpy()
y_confidence = y_confidence.numpy()
y_boxes[:, [0, 2]] = (y_boxes[:, [0, 2]] * image_wh[0] - padding[2]) / (
image_wh[0] - padding[2] - padding[3]) * image_size_old[0]
y_boxes[:, [1, 3]] = (y_boxes[:, [1, 3]] * image_wh[1] - padding[0]) / (
image_wh[1] - padding[0] - padding[1]) * image_size_old[1]
# 截取框超出图片部分
y_boxes[:, 0][y_boxes[:, 0] < 0] = 0
y_boxes[:, 1][y_boxes[:, 1] < 0] = 0
y_boxes[:, 2][y_boxes[:, 2] > image_size_old[0]] = image_size_old[0]
y_boxes[:, 3][y_boxes[:, 3] > image_size_old[1]] = image_size_old[1]
# 去掉无效框
y_mask = np.logical_and(y_boxes[:, 2] - y_boxes[:, 0] > 2,
y_boxes[:, 3] - y_boxes[:, 1] > 2)
y_boxes = y_boxes[y_mask]
y_classes_id = y_classes_id[y_mask]
y_scores = y_scores[y_mask]
y_classes = y_classes[y_mask]
y_confidence = y_confidence[y_mask]
y_boxes = y_boxes.astype(np.int32)
# print('y_boxes:', y_boxes.shape)
# print('y_classes_id:', y_classes_id.shape)
# print('y_scores:', y_scores.shape)
# print('y_classes:', y_classes.shape)
# print('y_confidence:', y_confidence.shape)
colors = []
for cr in range(0, 256, 32):
for cg in range(0, 256, 32):
for cb in range(0, 256, 32):
colors.append([cb, cg, cr])
colors = np.array(colors)
np.random.seed(11)
np.random.shuffle(colors)
np.random.seed(None)
result_img = img_old.copy()
for i in range(y_boxes.shape[0]):
# print('y_boxes:', y_boxes[i,:])
# print('y_classes_id:', y_classes_id[i])
# print('y_classes_name:', classes_name[y_classes_id[i]])
# print('y_scores:', y_scores[i])
# print('y_classes:', y_classes[i])
# print('y_confidence:', y_confidence[i])
cv2.rectangle(result_img,
tuple(y_boxes[i, 0:2]),
tuple(y_boxes[i, 2:4]),
colors[y_classes_id[i] % len(colors)].tolist(),
thickness=1)
cv2.putText(result_img, classes_name[y_classes_id[i]],
tuple(y_boxes[i, 0:2]), cv2.FONT_HERSHEY_COMPLEX, 0.6,
(0, 100, 0), 1)
cv2.putText(result_img, str(y_scores[i]),
tuple(y_boxes[i, 0:2] + (0, 20)), cv2.FONT_HERSHEY_COMPLEX,
0.6, (0, 0, 100), 1)
ImageHelper.showOpencvImage(result_img)
print('加载模型:{}'.format(last_model_path))
# model.summary()
file_list = ReadFileList(dataDir,
pattern=r'.*\.jpg',
select_sub_path=True,
is_full_path=False)
with open(outFile, 'w', encoding='utf-8') as f:
for image_path in file_list:
# 图片路径
full_path = os.path.join(dataDir, image_path)
img_old = ImageHelper.fileToOpencvImage(full_path)
# 缩放图片
img, _, padding = ImageHelper.opencvProportionalResize(img_old,
image_wh,
bg_color=(0, 0,
0))
# print('imgType:', type(img))
width, height = ImageHelper.opencvGetImageSize(img_old)
image_size_old = np.int32([width, height])
# 最后输出图片
predict_img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
# 调整参数范围
predict_img = predict_img.astype(np.float32)
predict_img = predict_img / 255
# 增加一个维度
predict_img = np.expand_dims(predict_img, 0)
y_boxes, y_classes_id, y_scores, y_classes, y_confidence = model.Predict(
predict_img,
confidence_thresh=0.5,