update

README.md

@@ -16,6 +16,9 @@
 * 通过运行read_datasests.py,可以对数据的标注信息进行解析可视化。      
 
 ### 2、静态手势识别数据集    
+
+ ![video_gesture](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/gesture.gif)  
+
 * 数据集来源3部分，且网上数据占绝大多数，具体：
 * 1）来源于网上数据并自制。
 * 2）来源于自己相机采集并自制。
@@ -29,23 +32,23 @@ kinect_leap_dataset 数据集来源于以下paper项目组的贡献。
 ```
 
 * 数据下载  
-  该项目数据集命名为：handpose_x_gesture_v1，[数据集下载地址(百度网盘 Password:  )]()  
+  该项目数据集命名为：handpose_x_gesture_v1，[数据集下载地址(百度网盘 Password: psi1 )](https://pan.baidu.com/s/10jN4DOYcu9Pfv6gd5c3y4A)  
   数据集共 2850 个样本，分为 14 类。  
 * 手势定义   
 
- ![one](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/one.jpg) ：000-one  
- ![five](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/five.jpg) ：001-five  
- ![fist](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/fist.jpg) ：002-fist  
- ![ok](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/ok.jpg) ：003-ok  
- ![heartSingle](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/heartSingle.jpg) ：004-heartSingle  
- ![yearh](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/yearh.jpg) ：005-yearh  
- ![three](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/three.jpg) ：006-three  
+ ![one](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/one.jpg) ：000-one  |
+ ![five](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/five.jpg) ：001-five  |
+ ![fist](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/fist.jpg) ：002-fist  |
+ ![ok](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/ok.jpg) ：003-ok   
+ ![heartSingle](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/heartSingle.jpg) ：004-heartSingle  |
+ ![yearh](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/yearh.jpg) ：005-yearh  |
+ ![three](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/three.jpg) ：006-three  |
  ![four](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/four.jpg) ：007-four  
- ![six](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/six.jpg) ：008-six  
- ![Iloveyou](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/Iloveyou.jpg) ：009-Iloveyou  
- ![gun](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/gun.jpg) ：010-gun  
+ ![six](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/six.jpg) ：008-six  |
+ ![Iloveyou](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/Iloveyou.jpg) ：009-Iloveyou  |
+ ![gun](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/gun.jpg) ：010-gun  |
  ![thumbUp](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/thumbUp.jpg) ：011-thumbUp  
- ![nine](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/nine.jpg) ：012-nine  
+ ![nine](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/nine.jpg) ：012-nine  |
  ![pink](https://codechina.csdn.net/EricLee/classification/-/raw/master/samples/pink.jpg) ：013-pink  
 
 ## 预训练模型    

inference.py

@@ -41,7 +41,7 @@ def get_xml_msg(path):
 if __name__ == "__main__":
 
     parser = argparse.ArgumentParser(description=' Project Classification top1 Test')
-    parser.add_argument('--test_model', type=str, default = './model_exp/2021-03-08_01-16-46/resnet_34-192_epoch-176.pth',
+    parser.add_argument('--test_model', type=str, default = './model_exp/2021-03-08_02-38-39/resnet_34-192_epoch-10.pth',
         help = 'test_model') # 模型路径
     parser.add_argument('--model', type=str, default = 'resnet_34',
         help = 'model : resnet_18,resnet_34,resnet_50,resnet_101,resnet_152') # 模型类型

train.py

@@ -230,7 +230,7 @@ def trainer(ops,f_log):
                     torch.save(model_.state_dict(), ops.model_exp + 'latest.pth')
             # 每间隔 5 个 epoch 进行模型保存
             if (epoch%5) == 0 and (epoch > 9):
-                torch.save(model_.state_dict(), ops.model_exp + '{}-{}_epoch-{}.pth'.format(ops.model,ops.img_size[0],epoch))
+                torch.save(model_.state_dict(), ops.model_exp + '{}-size-{}_epoch-{}.pth'.format(ops.model,ops.img_size[0],epoch))
 
             if len(val_split) > 0 and (epoch%ops.test_interval==0): # test
 
@@ -291,7 +291,7 @@ if __name__ == "__main__":
         help = 'init_learningRate') # 初始化学习率
     parser.add_argument('--lr_decay', type=float, default = 0.96,
         help = 'learningRate_decay') # 学习率权重衰减率
-    parser.add_argument('--weight_decay', type=float, default = 1e-8,
+    parser.add_argument('--weight_decay', type=float, default = 1e-6,
         help = 'weight_decay') # 优化器正则损失权重
     parser.add_argument('--batch_size', type=int, default = 32,
         help = 'batch_size') # 训练每批次图像数量