编写readme

LPRNet_ORT_infer.py

@@ -3,6 +3,7 @@ import cv2
 import numpy as np
 import argparse
 import os
+import time
 
 print('Runing Based On:', ort.get_device())
 
@@ -50,12 +51,19 @@ def LPRNetInference(args):
 
     if os.path.isdir(args.imgpath):
         images = os.listdir(args.imgpath)
+        count = 0
+        time1 = time.perf_counter()
         for image in images:
             img = LPRNetPreprocess(os.path.join(args.imgpath, image))
             intput = sess.get_inputs()[0].shape
             preb = sess.run(None, input_feed={sess.get_inputs()[0].name: img})[0]
             result = LPRNetPostprocess(preb)
+            if result == image[:-4]:
+                count += 1
             print('Inference Result:', result)
+        time2 = time.perf_counter()
+        print('accuracy rate:', count / len(images))
+        print('average time', (time2 - time1)/count*1000)        
     else:
         img = LPRNetPreprocess(args.imgpath)
         intput = sess.get_inputs()[0].shape

LPRNet_migraphx_infer.py

@@ -7,6 +7,7 @@ import numpy as np
 import migraphx
 import argparse
 import os
+import time
 
 CHARS = ['京', '沪', '津', '渝', '冀', '晋', '蒙', '辽', '吉', '黑',
          '苏', '浙', '皖', '闽', '赣', '鲁', '豫', '鄂', '湘', '粤',
@@ -56,14 +57,21 @@ def LPRNetInference(args):
 
     if os.path.isdir(args.imgpath):
         images = os.listdir(args.imgpath)
+        count = 0
+        time1 = time.perf_counter()
         for image in images:
             img = LPRNetPreprocess(os.path.join(args.imgpath, image))
-            inputName=model.get_parameter_names()[0]
-            inputShape=model.get_parameter_shapes()[inputName].lens()
+            inputName = model.get_parameter_names()[0]
+            inputShape = model.get_parameter_shapes()[inputName].lens()
             # print("inputName:{0} \ninputShape:{1}".format(inputName,inputShape))
             results = model.run({inputName: migraphx.argument(img)})
             result = LPRNetPostprocess(np.array(results[0]))
+            if result == image[:-4]:
+                count += 1
             print('Inference Result:', result)
+        time2 = time.perf_counter()
+        print('accuracy rate:', count / len(images))
+        print('average time', (time2 - time1)/count*1000) 
     else:
         img = LPRNetPreprocess(args.imgpath)
         inputName=model.get_parameter_names()[0]
@@ -75,8 +83,8 @@ def LPRNetInference(args):
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='parameters to vaildate net')
-    parser.add_argument('--model', default='model/LPRNet.mxr', help='model path to inference')
-    parser.add_argument('--imgpath', default='imgs/川JK0707.jpg', help='the image path')
+    parser.add_argument('--model', default='model/LPRNet.onnx', help='model path to inference')
+    parser.add_argument('--imgpath', default='imgs/京PL3N67.jpg', help='the image path')
     parser.add_argument('--savepath', default='model/LPRNet.mxr', help='mxr model save path and name')
     args = parser.parse_args()
 

README.md

@@ -42,9 +42,9 @@ LPRNet_migraphx_infer.py是基于Migraphx的推理脚本，使用需安装好Mig
 ## 性能和准确率数据
 测试数据使用的是[LPRNet_Pytorch](https://github.com/sirius-ai/LPRNet_Pytorch/tree/master/data/test)
 
-| size | personal test imgs(%) | DCU |
-| :------: | :------: | :------: |
-| 1.7M | 96.0+ | |
+| size | personal test imgs(%) | DCU(pth/ms) |DCU(onnx/ms) | DCU(mxr/ms)|
+| :------: | :------: | :------: |:------: |:------: |
+| 1.7M | 96.0+ | 52 | 51 | |
 ## 参考
 * [LPRNet: License Plate Recognition via Deep Neural Networks](https://arxiv.org/pdf/1806.10447v1.pdf)
 * [LPRNet_Pytorch](https://github.com/sirius-ai/LPRNet_Pytorch)

test.py

@@ -5,6 +5,7 @@ import torch
 import numpy as np
 from lprnet import build_lprnet
 from load_data import CHARS
+import time
 
 def infer(args, image, model):
     img = cv2.imread(image)
@@ -44,15 +45,18 @@ def validation(args):
     if os.path.isdir(args.imgpath):
         images = os.listdir(args.imgpath)
         count = 0
+        time1 = time.perf_counter()
         for image in images:
-            res = infer(args, os.path.join(args.imgpath, image), model)
-            if res == image[:-4]:
+            result = infer(args, os.path.join(args.imgpath, image), model)
+            if result == image[:-4]:
                 count += 1
-            print('Image: ' + image + ' recongise result: '+ res)
-        print('acc rate:', count / len(images))
+            print('Image: ' + image + ' recongise result: '+ result)
+        time2 = time.perf_counter()
+        print('accuracy rate:', count / len(images))
+        print('average time', (time2 - time1)/count*1000)
     else:
-        res = infer(args, args.imgpath, model)
-        print('Image: ' + args.imgpath + ' recongise result: '+ res)
+        result = infer(args, args.imgpath, model)
+        print('Image: ' + args.imgpath + ' recongise result: '+ result)
     
     if args.export_onnx:
         print('export pytroch model to onnx model...')
@@ -71,8 +75,8 @@ def validation(args):
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='parameters to vaildate net')
-    # parser.add_argument('--model', default='model/lprnet.pth', help='model path to vaildate')
-    parser.add_argument('--model', default='weights/Final_LPRNet_model.pth', help='model path to vaildate')
+    parser.add_argument('--model', default='model/lprnet.pth', help='model path to vaildate')
+    # parser.add_argument('--model', default='weights/Final_LPRNet_model.pth', help='model path to vaildate')
     # parser.add_argument('--imgpath', default='imgs/川JK0707.jpg', help='the image path')
     parser.add_argument('--imgpath', default='/code/lpr_ori/data/test', help='the image path')
     parser.add_argument('--device', default='cuda', help='Use cuda to vaildate model')