handwriting/paddle-handwrite/handwriting_socket.py

import socketserver
import struct
import base64
import numpy as np
import cv2
import time
from PIL import Image

from paddle.inference import Config
from paddle.inference import create_predictor

from handwriting_util import ResizeImage,CropImage,ToTensor,NormalizeImage

# 分包接收长度
BUFFER_SIZE = 1024

# paddle模型文件
MODEL_FILE = "output_2/inference/inference.pdmodel"
MODEL_PARAM = "output_2/inference/inference.pdiparams"
# 分配显存
GPU_MEM = 8000

# 前k个候选
top_k = 10

# 测试文件
img_file = "data/test/test_deal.png"
#标签文件
label_file = "data/trade/labels.txt"

# 建立Paddle预测器
def create_paddle_perdictor():
    config=Config(MODEL_FILE, MODEL_PARAM)
    config.enable_use_gpu(GPU_MEM, 0)
    config.set_cpu_math_library_num_threads(10)
    config.disable_glog_info()
    config.switch_ir_optim(True)
    config.enable_memory_optim()
    # use zero copy
    config.switch_use_feed_fetch_ops(False)
    predictor = create_predictor(config)
    return predictor

# 图像预处理，转为Tensor张量
def preprocess(img):
    resize_op = ResizeImage(resize_short=256)
    img = resize_op(img)
    crop_op = CropImage(size=(224, 224))
    img = crop_op(img)
    img_mean = [0.485, 0.456, 0.406]
    img_std = [0.229, 0.224, 0.225]
    img_scale = 1.0 / 255.0
    normalize_op = NormalizeImage(
        scale=img_scale, mean=img_mean, std=img_std)
    img = normalize_op(img)
    tensor_op = ToTensor()
    img = tensor_op(img)
    return img

# 结果分析处理
def postprocess(output):
    output = output.flatten()
    classes = np.argpartition(output, -top_k)[-top_k:]
    classes = classes[np.argsort(-output[classes])]
    scores = output[classes]
    return classes, scores

# 读取标签文件
def readLabels():
    index = 0
    labels=[]
    with open(label_file) as file_obj:
        for line in file_obj:
            labels.append(line.strip())
            index=index+1
    return labels

# 减少图像白边
def cutImg(img):
    dst = 255- img
    gray = cv2.cvtColor(dst,cv2.COLOR_BGR2GRAY)  
    ret, binary = cv2.threshold(gray,127,255,cv2.THRESH_BINARY)  
    
    contours, hierarchy = cv2.findContours(binary,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE)

    minx = 65523
    miny = 65523
    maxx = 0
    maxy = 0

    for contour in contours:
        boundRect = cv2.boundingRect(contour)
        (x,y,w,h)=boundRect
        if x<minx:
            minx=x
        if y<miny:
            miny=y
        if x+w>maxx:
            maxx=x+w
        if y+h>maxy:
            maxy=y+h
    cutw=maxx-minx
    cuth=maxy-miny
    img = img[miny:maxy,minx:maxx]
    return img

all_label = readLabels()
predictor = create_paddle_perdictor()

class MyTCPHandler(socketserver.BaseRequestHandler):
    def handle(self):
        try:
            
            self.data_len=self.request.recv(4)
            (byte_len,) = struct.unpack(">I", self.data_len)
            print("{} received byte len:".format(self.client_address),byte_len)
            # 分包接收
            content = b''
            while byte_len >0:
                if byte_len>=BUFFER_SIZE:
                    tmp = self.request.recv(BUFFER_SIZE)
                    content += tmp
                    byte_len -= BUFFER_SIZE
                else:
                    tmp = self.request.recv(byte_len)
                    content += tmp
                    byte_len = 0
                    break
            
            base64content = content
            # print("{} received image:".format(self.client_address), base64content)
            
            # 进行预测
            input_names = predictor.get_input_names()
            input_tensor = predictor.get_input_handle(input_names[0])
            output_names = predictor.get_output_names()
            output_tensor = predictor.get_output_handle(output_names[0])
            imgString = base64.b64decode(base64content)
            nparr = np.fromstring(imgString,np.uint8)  
            img_recv = cv2.imdecode(nparr,cv2.IMREAD_UNCHANGED)
            

            #make mask of where the transparent bits are
            trans_mask = img_recv[:,:,3] == 0
            #replace areas of transparency with white and not transparent
            img_recv[trans_mask] = [255, 255, 255, 255]
            img_save = cv2.cvtColor(img_recv, cv2.COLOR_BGRA2BGR)
            # cv2.imwrite("data/test/receive.png", img_save)
            img_save = cutImg(img_save)
            img = cv2.resize(img_save,(64,64))
            inputs = preprocess(img)
            inputs = np.expand_dims(
                inputs, axis=0).repeat(
                    1, axis=0).copy()
            input_tensor.copy_from_cpu(inputs)
            predictor.run()
            output = output_tensor.copy_to_cpu()
            classes, scores = postprocess(output)
            print(classes)
            result=""
            for index in classes:
                label=all_label[index]
                print(label)
                result = result+label
            print(result)
            self.request.sendall(bytes(result, encoding = "utf8"))
                
        except Exception as e:

            print("出现如下异常%s"%e)
        finally:
            print(self.client_address,"连接断开")

            self.request.close()
    def setup(self):
        print("before handle,连接建立：",self.client_address)
    def finish(self):
        print("finish run  after handle")


if __name__=="__main__":
    HOST,PORT = "0.0.0.0",8123
    server=socketserver.TCPServer((HOST,PORT),MyTCPHandler)
    print('启动端口8123')
    server.serve_forever()