-
Notifications
You must be signed in to change notification settings - Fork 2.9k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update SSD documentation #987
Conversation
fluid/object_detection/train.py
Outdated
model_path = os.path.join(model_save_dir, str(pass_id)) | ||
print 'save models to %s' % (model_path) | ||
fluid.io.save_persistables(exe, model_path) | ||
|
||
|
||
def parallel_exe(args, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
parallel_exe -> train
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
revised
fluid/object_detection/README_cn.md
Outdated
训练数据的读取行为定义在 `reader.py` 中。所有的图片都会被resize到300x300。在训练时,图片还会被随机扰动、扩张、裁剪和翻转: | ||
- 扰动: 扰动图片亮度、对比度、饱和度和hue。 | ||
- 扩张: 将原始图片放进一张使用像素均值填充的扩张图中,之后对此图进行resize则相当于缩小了原图。 | ||
- 裁剪: 对图片进行不同大小、比例和IOU的裁剪。 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这个描述看不太懂,需要修改下。
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
revised
fluid/object_detection/README_cn.md
Outdated
@@ -70,7 +74,13 @@ cd data/coco | |||
python train.py --help | |||
``` | |||
|
|||
我们使用了 RMSProp 优化算法来训练 MobileNet-SSD,batch大小为64,权重衰减系数为0.00005,初始学习率为 0.001,并且在第40、60、80、100 轮时使用 0.5, 0.25, 0.1, 0.01乘子进行学习率衰减。在120轮训练后,11point评价标准下的mAP为XXX%。 | |||
训练数据的读取行为定义在 `reader.py` 中。所有的图片都会被resize到300x300。在训练时,图片还会被随机扰动、扩张、裁剪和翻转: | |||
- 扰动: 扰动图片亮度、对比度、饱和度和hue。 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
hue换成中文?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
revised
fluid/object_detection/README_cn.md
Outdated
@@ -70,7 +74,13 @@ cd data/coco | |||
python train.py --help | |||
``` | |||
|
|||
我们使用了 RMSProp 优化算法来训练 MobileNet-SSD,batch大小为64,权重衰减系数为0.00005,初始学习率为 0.001,并且在第40、60、80、100 轮时使用 0.5, 0.25, 0.1, 0.01乘子进行学习率衰减。在120轮训练后,11point评价标准下的mAP为XXX%。 | |||
训练数据的读取行为定义在 `reader.py` 中。所有的图片都会被resize到300x300。在训练时,图片还会被随机扰动、扩张、裁剪和翻转: |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
resize -> 缩放
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
revised
fluid/object_detection/README_cn.md
Outdated
- 扰动: 扰动图片亮度、对比度、饱和度和色相。 | ||
- 扩张: 将原始图片放进一张使用像素均值填充的扩张图中,之后对此图进行缩放则相当于缩放了原图。 | ||
- 翻转: 水平翻转。 | ||
- 采样: 依据缩放比例、长宽比例,生成若干候选坐标,再依据这些候选坐标和真实坐标的面积交并比(IOU)采样出若干坐标。 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
随机裁剪: 根据缩放比例、长宽比例两个参数生成若干候选框,再依据这些候选框和标注框的面积交并比(IoU)挑选出符合要求的裁剪结果。 这么写是不是好一点 替换掉原来的采样那一行
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
revised, 使用"裁剪"而不是"随机裁剪",因为扰动,扩展,翻转,裁剪都是随机,要么统一加:“随机”,要么统一不加。因为统一加太冗余,所以统一不加
fluid/object_detection/README_cn.md
Outdated
我们使用了 RMSProp 优化算法来训练 MobileNet-SSD,batch大小为64,权重衰减系数为0.00005,初始学习率为 0.001,并且在第40、60、80、100 轮时使用 0.5, 0.25, 0.1, 0.01乘子进行学习率衰减。在120轮训练后,11point评价标准下的mAP为XXX%。 | ||
数据的读取行为定义在 `reader.py` 中,所有的图片都会被缩放到300x300。在训练时,数据还会进行图片增强和标签增强,图片增强包括对图片本身的随机扰动、扩张和翻转,标签增强包括采样: | ||
- 扰动: 扰动图片亮度、对比度、饱和度和色相。 | ||
- 扩张: 将原始图片放进一张使用像素均值填充的扩张图中,之后对此图进行缩放则相当于缩放了原图。 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
这里不用之后了吧 就写再对此图进行缩放是不是好一点。填均值这里要不直接写填0或者注明一下 后面会减掉均值?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
revised
No description provided.