def execute(self):
"""执行业务逻辑"""
utils.info(
'API REQUEST INFO[' + self.request.path + '][' +
self.request.method + '][' + self.request.remote_ip + '][' +
str(self.request.arguments) + ']', ApiImageDigit)
img_file = self.get_argument('img_file', '')
model_path = self.get_argument('model_path', '')
if img_file == '':
return {'code': 2, 'msg': 'img_file不能为空'}
if model_path == '':
model_path = MODEL_PATH
res = {}
try:
ret, msg, res = infer.infer(img_file, model_path)
if ret != 0:
utils.error('execute fail [' + img_file + '] ' + msg,
ApiImageDigit)
return {'code': 4, 'msg': '查询失败'}
except:
utils.error('execute fail [' + img_file + '] ' + utils.get_trace(),
ApiImageDigit)
return {'code': 5, 'msg': '查询失败'}
# 组织返回格式
return {'code': 0, 'msg': 'success', 'data': res}
def post(self):
"""post请求处理"""
try:
result = self.execute()
except:
utils.error('execute fail ' + utils.get_trace(), ApiImageDigit)
result = {'code': 1, 'msg': '查询失败'}
utils.info('API RES[' + self.request.path + '][' + self.request.method + ']['
+ str(result['code']) + '][' + str(result['msg']) + ']', ApiImageDigit)
self.write(json.dumps(result))
def infer(img_file='', model_path='./'):
if img_file == '':
opts, args = getopt.getopt(sys.argv[1:], "p:", ["file_name="])
if len(args) == 0:
print(
"usage: python infer.py [file_name_name] \n\tpython infer.py infer_62.jpeg"
)
return 1, 'file_name is empty', {}
file_name = args[0]
img_file = './data/image/' + file_name
else:
file_name = img_file.split('/')[-1]
# 是否使用GPU
use_cuda = False # set to True if training with GPU
place = fluid.CUDAPlace(0) if use_cuda else fluid.CPUPlace()
params_dirname = model_path + "/model/" # 模型参数保存目录
height = width = 28
# 加载测试数据
imgs = [] #使用多种不同的预处理方法
imgs_weight = [1, 0.99, 0.99] #不同预处理方法的结果权重
try: #测试集图片
imgs.append(utils.load_image(img_file, height, width))
except:
utils.warning(utils.get_trace())
imgs.append([])
try: #白纸手写照片
if len(file_name.split('_')[1].split('.')[0]) >= 2 and int(
file_name.split('_')[1][1:2]) > 0:
imgs_weight[1] = 5
imgs.append(
utils.load_image(img_file,
height,
width,
rotate=0,
sobel=True,
save_resize=True,
ksize=5,
dilate=1))
except:
utils.warning(utils.get_trace(), 'infer')
imgs.append([])
try: #黑纸粗笔写照片
imgs.append(
utils.load_image(img_file,
height,
width,
rotate=0,
sobel=True,
save_resize=True,
ksize=3,
dilate=6,
erode=1))
except:
utils.warning(utils.get_trace(), 'infer')
imgs.append([])
# 使用保存的模型参数+测试图片进行预测
inferencer = Inferencer(
# infer_func=softmax_regression, # uncomment for softmax regression
# infer_func=multilayer_perceptron, # uncomment for MLP
infer_func=utils.convolutional_neural_network, # uncomment for LeNet5
param_path=params_dirname,
place=place)
results = []
results_sum = numpy.ndarray([])
numpy.set_printoptions(precision=2)
for i in xrange(len(imgs)):
if len(imgs[i]) == 0: continue
result = inferencer.infer({'img': imgs[i]}) #此输入img的各label概率
result = numpy.where(result[0][0] > 0.01, result[0][0],
0) #概率<0.1%的直接设置为0
print(result)
results.append(result)
print(numpy.argsort(result))
results_sum = results_sum + result * imgs_weight[i] #累加label下标概率
#print(imgs_weight)
#按概率加和排序
lab = numpy.argsort(
results_sum) # probs and lab are the results of one batch data
label = lab[-1] #概率倒排最后一个
weight = []
for result in results:
if numpy.argsort(result)[-1] == label:
weight = result
print("*label weight sort:")
print(results_sum)
print(lab)
print("*img: %s" % img_file)
print("*label: %d weight: %f" % (label, weight[label]))
return 0, '', {
'img': img_file,
'label': label,
'weight': list(weight.astype(str))
}