docs(mkdocs): 分别使用location dataset和voc 07数据集进行测试

Author: zjZSTU

docs/index.md

@@ -10,7 +10,9 @@
 4. 训练模型
 5. 计算`mAP`
 
-`50`轮训练完成后能够实现`97.31%`的`mAP`
+对于`Location DataSet`, `50`轮训练完成后能够实现`97.31% mAP`
+
+对于`VOC 07`，`50`轮训练完成后能够实现`xxx mAP`以及`xxx FPS`
 
 ## 相关链接
 

docs/log.md docs/location-log.mddocs/log.md renamed to docs/location-log.md

@@ -1,5 +1,5 @@
 
-# 日志
+# 定位数据集训练日志
 
 ## 训练参数
 

docs/voc-07-log.md

@@ -0,0 +1,272 @@
+
+# 07 VOC训练日志
+
+## 训练参数
+
+* `S=7, B=2, C=3`
+* 缩放至`(448, 448)`，进行数据标准化处理
+* 优化器：`SGD`，学习率`1e-3`，动量大小`0.9`
+* 衰减器：每隔`4`轮衰减`4%`，学习因子`0.96`
+
+## 检测结果
+
+```
+。。。
+```
+
+## 训练日志
+
+```
+$ python train.py 
+Epoch 0/49
+----------
+train Loss: 9.3544
+save model
+
+Epoch 1/49
+----------
+train Loss: 7.4459
+save model
+
+Epoch 2/49
+----------
+train Loss: 7.1511
+save model
+
+Epoch 3/49
+----------
+train Loss: 6.9450
+save model
+
+Epoch 4/49
+----------
+train Loss: 6.7656
+save model
+
+Epoch 5/49
+----------
+train Loss: 6.6077
+save model
+
+Epoch 6/49
+----------
+train Loss: 6.4264
+save model
+
+Epoch 7/49
+----------
+train Loss: 6.2750
+save model
+
+Epoch 8/49
+----------
+train Loss: 6.0318
+save model
+
+Epoch 9/49
+----------
+train Loss: 5.7777
+save model
+
+Epoch 10/49
+----------
+train Loss: 5.4760
+save model
+
+Epoch 11/49
+----------
+train Loss: 5.1784
+save model
+
+Epoch 12/49
+----------
+train Loss: 4.8067
+save model
+
+Epoch 13/49
+----------
+train Loss: 4.4603
+save model
+
+Epoch 14/49
+----------
+train Loss: 4.1291
+save model
+
+Epoch 15/49
+----------
+train Loss: 3.7810
+save model
+
+Epoch 16/49
+----------
+train Loss: 3.4259
+save model
+
+Epoch 17/49
+----------
+train Loss: 3.1166
+save model
+
+Epoch 18/49
+----------
+train Loss: 2.8041
+save model
+
+Epoch 19/49
+----------
+train Loss: 2.4853
+save model
+
+Epoch 20/49
+----------
+train Loss: 2.2107
+save model
+
+Epoch 21/49
+----------
+train Loss: 1.9424
+save model
+
+Epoch 22/49
+----------
+train Loss: 1.7307
+save model
+
+Epoch 23/49
+----------
+train Loss: 1.5518
+save model
+
+Epoch 24/49
+----------
+train Loss: 1.3712
+save model
+
+Epoch 25/49
+----------
+train Loss: 1.2174
+save model
+
+Epoch 26/49
+----------
+train Loss: 1.1142
+save model
+
+Epoch 27/49
+----------
+train Loss: 1.0305
+save model
+
+Epoch 28/49
+----------
+train Loss: 0.9270
+save model
+
+Epoch 29/49
+----------
+train Loss: 0.8465
+save model
+
+Epoch 30/49
+----------
+train Loss: 0.7739
+save model
+
+Epoch 31/49
+----------
+train Loss: 0.7228
+save model
+
+Epoch 32/49
+----------
+train Loss: 0.6765
+save model
+
+Epoch 33/49
+----------
+train Loss: 0.6264
+save model
+
+Epoch 34/49
+----------
+train Loss: 0.5932
+save model
+
+Epoch 35/49
+----------
+train Loss: 0.5646
+save model
+
+Epoch 36/49
+----------
+train Loss: 0.5359
+save model
+
+Epoch 37/49
+----------
+train Loss: 0.4941
+save model
+
+Epoch 38/49
+----------
+train Loss: 0.4682
+save model
+
+Epoch 39/49
+----------
+train Loss: 0.4482
+save model
+
+Epoch 40/49
+----------
+train Loss: 0.4276
+save model
+
+Epoch 41/49
+----------
+train Loss: 0.4048
+save model
+
+Epoch 42/49
+----------
+train Loss: 0.3874
+save model
+
+Epoch 43/49
+----------
+train Loss: 0.3759
+save model
+
+Epoch 44/49
+----------
+train Loss: 0.3607
+save model
+
+Epoch 45/49
+----------
+train Loss: 0.3415
+save model
+
+Epoch 46/49
+----------
+train Loss: 0.3327
+save model
+
+Epoch 47/49
+----------
+train Loss: 0.3222
+save model
+
+Epoch 48/49
+----------
+train Loss: 0.3141
+save model
+
+Epoch 49/49
+----------
+train Loss: 0.2978
+save model
+
+Training complete in 214m 30s
+```

docs/数据集.md

@@ -1,13 +1,20 @@
 
 # 数据集
 
-当前使用`3`类定位数据集，参考[[数据集]Image Localization Dataset](https://blog.zhujian.life/posts/a2d65e1.html)
+当前使用了两种数据集进行测试：
+
+* 定位数据集（`3`类）
+* `VOC 07`数据集（`20`类）
+
+## 定位数据集
+
+参考[[数据集]Image Localization Dataset](https://blog.zhujian.life/posts/a2d65e1.html)
 
 ```
 {'cucumber': 63, 'mushroom': 61, 'eggplant': 62}
 ```
 
-## 解析
+### 解析
 
 下载数据集后，解压到`py/data`目录，得到`training_images`，其格式如下：
 
@@ -43,4 +50,8 @@ $ python parse_location.py
     ├── cucumber_11.jpg
 。。。
 。。。
-```
+```
+
+## VOC 07
+
+。。。