def predict(configs, detection):
global predictor
width = configs['input_width']
height = configs['input_height']
image = read_image(configs)
input = preprocess_image(image, configs)
tensor = pm.PaddleTensor()
tensor.dtype = pm.PaddleDType.FLOAT32
tensor.shape = (1, 3, width, height)
tensor.data = pm.PaddleBuf(input)
paddle_data_feeds = [tensor]
print('prediction is running ...')
outputs = predictor.Run(paddle_data_feeds)
assert len(
outputs
) == 1, 'error numbers of tensor returned from Predictor.Run function !!!'
output = np.array(outputs[0], copy=False)
print('output:', output)
print('\nprediction result :')
print('\t nDim: ' + str(output.ndim))
print('\tShape: ' + str(output.shape))
print('\tDType: ' + str(output.dtype))
# print(output)
# print('')
if detection:
detect(output, configs)
else:
classify(output, configs)
def init_tensor(self, data_shape):
"""
初始化PaddleMobile模型输入数据张量
:param data_shape: 数据形状
:return: 数据张量
"""
tensor = pm.PaddleTensor()
tensor.dtype = pm.PaddleDType.FLOAT32
tensor.shape = data_shape
return tensor
def data_feed(self, data_shape):
'''
初始化PaddleMobile模型输入数据张量
参数：数据形状
返回：数据张量
'''
tensor = pm.PaddleTensor()
tensor.dtype = pm.PaddleDType.FLOAT32
tensor.shape = (data_shape)
return tensor
def tensor_deal(self, origin):
tensor_img = origin.resize((256, 256), Image.BILINEAR)
if tensor_img.mode != 'RGB':
tensor_img = tensor_img.convert('RGB')
tensor_img = np.array(tensor_img).astype('float32').transpose(
(2, 0, 1))
tensor_img -= 127.5
tensor_img *= 0.007843
tensor_img = tensor_img[np.newaxis, :]
tensor = pm.PaddleTensor()
tensor.dtype = pm.PaddleDType.FLOAT32
tensor.shape = (1, 3, 256, 256)
tensor.data = pm.PaddleBuf(tensor_img)
paddle_data_feeds = [tensor]
return paddle_data_feeds
predictor = load_model()
while True:
# 保存图像 调试用
select.select((video,), (), ())
image_data = video.read_and_queue()
frame = cv2.imdecode(np.frombuffer(image_data, dtype=np.uint8), cv2.IMREAD_COLOR)
cv2.imwrite("test.jpg", frame)
# 图像预处理
origin, img = dataset(video)
tensor_img = origin.resize((244, 244), Image.BILINEAR)
if tensor_img.mode != 'RGB':
tensor_img = tensor_img.convert('RGB')
tensor_img = np.array(tensor_img).astype('float32').transpose((2, 0, 1)) # HWC to CHW
tensor = pm.PaddleTensor()
tensor.dtype =pm.PaddleDType.FLOAT32
tensor.shape = (1,1,244,244)
tensor.data = pm.PaddleBuf(tensor_img)
paddle_data_feeds = [tensor]
# 将图像输入神经网络并获得输出
outputs = predictor.Run(paddle_data_feeds)
# 处理输出结果并返回最终预测值
output = np.array(outputs[0], copy = False)
# print(output)
number = np.argsort(output)
print("The number is ", number[0][-1])
def init_tensor(self, data_shape):
tensor = pm.PaddleTensor()
tensor.dtype = pm.PaddleDType.FLOAT32
tensor.shape = data_shape
return tensor