-
Notifications
You must be signed in to change notification settings - Fork 37
/
inference阶段代码复现环境执行顺序.txt
21 lines (17 loc) · 1.29 KB
/
inference阶段代码复现环境执行顺序.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
环境配置要求:
依赖:
Python 3.6
Python opencv
TensorFlow 1.8.0 至少大于等于1.6.0
本代码使用了tensorpack库 故需要安装tensorpack:pip install --upgrade git+https://github.com/tensorpack/tensorpack.git
安装pycocotools: pip install 'git+https://github.com/cocodataset/cocoapi.git#subdirectory=PythonAPI'
其他依赖:tqdm,numpy,matplotlib,pandas
操作顺序:
更改运行脚本train_11_5中的读取路径:
先更改67,68,72行的文件路径:intensity_path='to/your/intensity_files'
pts_path='to/your/pts_files'
upload_path='to/your/upload_csv_file'
打开终端进入配置好的环境再进入本复现代码文件夹:
输入命令:python train_11_5.py --predict /to/your/pts_files/ --load /your/path/to/FasterRCNN/train_log/maskrcnn1/model-1960000.data-00000-of-00001 --config MODE_MASK=False MODE_FPN=True BACKBONE.WEIGHTS=/your/path/to/FasterRCNN/ImageNet-R101-AlignPadding.npz
示例:python train_11_5.py --predict /media/wrc/DCFfusai/dianyun/pts_2/ --load /media/wrc/CCFtijiao/FasterRCNN/train_log/maskrcnn1/model-1960000.data-00000-of-00001 --config MODE_MASK=False MODE_FPN=True BACKBONE.WEIGHTS=/media/wrc/CCFtijiao/FasterRCNN/ImageNet-R101-AlignPadding
在文件夹中的upload文件夹中得到结果文件