《k230-AI_DEMO》车牌识别

《k230-AI

'''
实验名称：车牌识别
实验平台：01Studio CanMV K230
教程：wiki.01studio.cc
'''from libs.PipeLine import PipeLine, ScopedTiming
from libs.AIBase import AIBase
from libs.AI2D import Ai2d
import os
import ujson
from media.media import *
from time import *
import nncase_runtime as nn
import ulab.numpy as np
import time
import image
import aidemo
import random
import gc
import sys# 自定义车牌检测类
class LicenceDetectionApp(AIBase):# 初始化函数，设置车牌检测应用的参数def __init__(self, kmodel_path, model_input_size, confidence_threshold=0.5, nms_threshold=0.2, rgb888p_size=[224,224], display_size=[1920,1080], debug_mode=0):super().__init__(kmodel_path, model_input_size, rgb888p_size, debug_mode)  # 调用基类的初始化函数self.kmodel_path = kmodel_path  # 模型路径# 模型输入分辨率self.model_input_size = model_input_size# 分类阈值self.confidence_threshold = confidence_thresholdself.nms_threshold = nms_threshold# sensor给到AI的图像分辨率self.rgb888p_size = [ALIGN_UP(rgb888p_size[0], 16), rgb888p_size[1]]# 显示分辨率self.display_size = [ALIGN_UP(display_size[0], 16), display_size[1]]self.debug_mode = debug_mode# Ai2d实例，用于实现模型预处理self.ai2d = Ai2d(debug_mode)# 设置Ai2d的输入输出格式和类型self.ai2d.set_ai2d_dtype(nn.ai2d_format.NCHW_FMT, nn.ai2d_format.NCHW_FMT, np.uint8, np.uint8)# 配置预处理操作，这里使用了pad和resize，Ai2d支持crop/shift/pad/resize/affinedef config_preprocess(self, input_image_size=None):with ScopedTiming("set preprocess config", self.debug_mode > 0):# 初始化ai2d预处理配置，默认为sensor给到AI的尺寸，可以通过设置input_image_size自行修改输入尺寸ai2d_input_size = input_image_size if input_image_size else self.rgb888p_sizeself.ai2d.resize(nn.interp_method.tf_bilinear, nn.interp_mode.half_pixel)self.ai2d.build([1,3,ai2d_input_size[1],ai2d_input_size[0]],[1,3,self.model_input_size[1],self.model_input_size[0]])# 自定义当前任务的后处理def postprocess(self, results):with ScopedTiming("postprocess", self.debug_mode > 0):# 对检测结果进行后处理det_res = aidemo.licence_det_postprocess(results, [self.rgb888p_size[1], self.rgb888p_size[0]], self.model_input_size, self.confidence_threshold, self.nms_threshold)return det_res# 自定义车牌识别任务类
class LicenceRecognitionApp(AIBase):def __init__(self,kmodel_path,model_input_size,rgb888p_size=[1920,1080],display_size=[1920,1080],debug_mode=0):super().__init__(kmodel_path,model_input_size,rgb888p_size,debug_mode)# kmodel路径self.kmodel_path=kmodel_path# 检测模型输入分辨率self.model_input_size=model_input_size# sensor给到AI的图像分辨率，宽16字节对齐self.rgb888p_size=[ALIGN_UP(rgb888p_size[0],16),rgb888p_size[1]]# 视频输出VO分辨率，宽16字节对齐self.display_size=[ALIGN_UP(display_size[0],16),display_size[1]]# debug模式self.debug_mode=debug_mode# 车牌字符字典self.dict_rec = ["挂", "使", "领", "澳", "港", "皖", "沪", "津", "渝", "冀", "晋", "蒙", "辽", "吉", "黑", "苏", "浙", "京", "闽", "赣", "鲁", "豫", "鄂", "湘", "粤", "桂", "琼", "川", "贵", "云", "藏", "陕", "甘", "青", "宁", "新", "警", "学", "0", "1", "2", "3", "4", "5", "6", "7", "8", "9", "A", "B", "C", "D", "E", "F", "G", "H", "J", "K", "L", "M", "N", "P", "Q", "R", "S", "T", "U", "V", "W", "X", "Y", "Z", "_", "-"]self.dict_size = len(self.dict_rec)self.ai2d=Ai2d(debug_mode)self.ai2d.set_ai2d_dtype(nn.ai2d_format.NCHW_FMT,nn.ai2d_format.NCHW_FMT,np.uint8, np.uint8)# 配置预处理操作，这里使用了resize，Ai2d支持crop/shift/pad/resize/affinedef config_preprocess(self,input_image_size=None):with ScopedTiming("set preprocess config",self.debug_mode > 0):ai2d_input_size=input_image_size if input_image_size else self.rgb888p_sizeself.ai2d.resize(nn.interp_method.tf_bilinear, nn.interp_mode.half_pixel)self.ai2d.build([1,3,ai2d_input_size[1],ai2d_input_size[0]],[1,3,self.model_input_size[1],self.model_input_size[0]])# 自定义后处理，results是模型输出的array列表def postprocess(self,results):with ScopedTiming("postprocess",self.debug_mode > 0):output_data=results[0].reshape((-1,self.dict_size))max_indices = np.argmax(output_data, axis=1)result_str = ""for i in range(max_indices.shape[0]):index = max_indices[i]if index > 0 and (i == 0 or index != max_indices[i - 1]):result_str += self.dict_rec[index - 1]return result_str# 车牌识别任务类
class LicenceRec:def __init__(self,licence_det_kmodel,licence_rec_kmodel,det_input_size,rec_input_size,confidence_threshold=0.25,nms_threshold=0.3,rgb888p_size=[1920,1080],display_size=[1920,1080],debug_mode=0):# 车牌检测模型路径self.licence_det_kmodel=licence_det_kmodel# 车牌识别模型路径self.licence_rec_kmodel=licence_rec_kmodel# 人脸检测模型输入分辨率self.det_input_size=det_input_size# 人脸姿态模型输入分辨率self.rec_input_size=rec_input_size# 置信度阈值self.confidence_threshold=confidence_threshold# nms阈值self.nms_threshold=nms_threshold# sensor给到AI的图像分辨率，宽16字节对齐self.rgb888p_size=[ALIGN_UP(rgb888p_size[0],16),rgb888p_size[1]]# 视频输出VO分辨率，宽16字节对齐self.display_size=[ALIGN_UP(display_size[0],16),display_size[1]]# debug_mode模式self.debug_mode=debug_modeself.licence_det=LicenceDetectionApp(self.licence_det_kmodel,model_input_size=self.det_input_size,confidence_threshold=self.confidence_threshold,nms_threshold=self.nms_threshold,rgb888p_size=self.rgb888p_size,display_size=self.display_size,debug_mode=0)self.licence_rec=LicenceRecognitionApp(self.licence_rec_kmodel,model_input_size=self.rec_input_size,rgb888p_size=self.rgb888p_size)self.licence_det.config_preprocess()# run函数def run(self,input_np):# 执行车牌检测det_boxes=self.licence_det.run(input_np)# 将车牌部分抠出来imgs_array_boxes = aidemo.ocr_rec_preprocess(input_np,[self.rgb888p_size[1],self.rgb888p_size[0]],det_boxes)imgs_array = imgs_array_boxes[0]boxes = imgs_array_boxes[1]rec_res = []for img_array in imgs_array:# 对每一个检测到的车牌进行识别self.licence_rec.config_preprocess(input_image_size=[img_array.shape[3],img_array.shape[2]])licence_str=self.licence_rec.run(img_array)rec_res.append(licence_str)gc.collect()return det_boxes,rec_res# 绘制车牌检测识别效果def draw_result(self,pl,det_res,rec_res):pl.osd_img.clear()if det_res:point_8 = np.zeros((8),dtype=np.int16)for det_index in range(len(det_res)):for i in range(4):x = det_res[det_index][i * 2 + 0]/self.rgb888p_size[0]*self.display_size[0]y = det_res[det_index][i * 2 + 1]/self.rgb888p_size[1]*self.display_size[1]point_8[i * 2 + 0] = int(x)point_8[i * 2 + 1] = int(y)for i in range(4):pl.osd_img.draw_line(point_8[i * 2 + 0],point_8[i * 2 + 1],point_8[(i+1) % 4 * 2 + 0],point_8[(i+1) % 4 * 2 + 1],color=(255, 0, 255, 0),thickness=4)pl.osd_img.draw_string_advanced( point_8[6], point_8[7] + 20, 40,rec_res[det_index] , color=(255,255,153,18))if __name__=="__main__":# 显示模式，默认"hdmi",可以选择"hdmi"和"lcd"display_mode="lcd"if display_mode=="hdmi":display_size=[1920,1080]else:display_size=[800,480]# 车牌检测模型路径licence_det_kmodel_path="/sdcard/examples/kmodel/LPD_640.kmodel"# 车牌识别模型路径licence_rec_kmodel_path="/sdcard/examples/kmodel/licence_reco.kmodel"# 其它参数rgb888p_size=[640,360]licence_det_input_size=[640,640]licence_rec_input_size=[220,32]confidence_threshold=0.2nms_threshold=0.2# 初始化PipeLine，只关注传给AI的图像分辨率，显示的分辨率pl=PipeLine(rgb888p_size=rgb888p_size,display_size=display_size,display_mode=display_mode)pl.create()lr=LicenceRec(licence_det_kmodel_path,licence_rec_kmodel_path,det_input_size=licence_det_input_size,rec_input_size=licence_rec_input_size,confidence_threshold=confidence_threshold,nms_threshold=nms_threshold,rgb888p_size=rgb888p_size,display_size=display_size)clock = time.clock()while True:clock.tick()img = pl.get_frame()  # 获取当前帧det_res, rec_res = lr.run(img)  # 推理当前帧lr.draw_result(pl, det_res, rec_res)  # 绘制当前帧推理结果# 清晰打印车牌识别结果if rec_res:print("识别到的车牌有:")for l in rec_res:print(l)if l == '粤BF12345':print(666)else:print("未识别到车牌。")pl.show_image()  # 展示推理结果gc.collect()print(clock.fps())  # 打印帧率