Jackie-Chou
commented
5 years ago Hi, sorry for the unclear explanation of the parameters. Here’s more detailed interpretation:
data_dir: this is the root directory of the input data/images. save_dir: this is the root directory to save out files. These two parameters are a little bit tricky. As you may see in match.py, I use string.replace method to produce the paths for output files from input image paths. The data_dir is generally a common prefix for all input image paths, and it is replaced by save_dir prefix to make the output file paths.
tag: this is just a small token used to identify one particular execution of match.py. As you can see in match.py Line-69,70, tag is inserted into the submission directory paths and create the according directories.
To be more clear, here’s one specific example. As you may see in data/list/train_all.txt my input image paths. Now I set data_dir to “/scratch/xz/MC-CNN-python/data/MiddEval3” and save_dir to “/scratch/xz/MC-CNN-python/data/result”, and I set tag to “abc”.
Then in match.py line-69 and line-70 we have: save_res_dir = /scratch/xz/MC-CNN-python/data/result/submit_abc save_img_dir = /scratch/xz/MC-CNN-python/data/result/submit_abc_imgs
Assume now the input left image is from path “/scratch/xz/MC-CNN-python/data/MiddEval3/trainingH/Playtable/im0.png” Then as line-99 to line-103 we have: res_dir = “/scratch/xz/MC-CNN-python/data/result/submit_abc/trainingH/Playtable” img_dir = “/scratch/xz/MC-CNN-python/data/result/submit_abc_imgs/trainingH/Playtable”
Then as line-108 to line-110 we have: out_path = “/scratch/xz/MC-CNN-python/data/result/submit_abc/trainingH/Playtable/disp0MCCNN.pfm” out_time_path = “/scratch/xz/MC-CNN-python/data/result/submit_abc/trainingH/Playtable/timeMCCNN.txt” out_img_path = “/scratch/xz/MC-CNN-python/data/result/submit_abc/trainingH/Playtable/disp0MCCNN.pgm”
The above three files are the actual output files. I recommend you to read code in match.py from line-64 to line-100 about details of these paths.
start, end: these are used to control match.py to only process a subset of input images. The subset starts from index “start” to index “end”. For example, setting start=1 and end=2 means to only process image No.1 and No.2. You can parallelize processing for different images by dividing all images into multiple subsets and run them simultaneously.
Hope this can be helpful. Best regards.
Jackie Chou
发送自 Windows 10 版邮件应用
发件人: hurricane2018 发送时间: 2018年12月18日 20:21 收件人: Jackie-Chou/MC-CNN-python 抄送: Subscribed 主题: [Jackie-Chou/MC-CNN-python] Can you give more details about theparameters ? (#2)
Hi, thank you for your wonderful work and code. However, when I run the "python match.py", the error occurs. Could you give more details about the parameters including data_dir, save_dir, t, s, and e?
parser.add_argument("--data_dir", type=str, required=True, help="path to root dir to data.") parser.add_argument("--save_dir", type=str, required=True, help="path to root dir to save results") parser.add_argument("-t", "--tag", type=str, required=True, help="tag used to indicate one run") parser.add_argument("-s", "--start", type=int, required=True, help="index of first image to do matching, this is used for parallel matching of different images") parser.add_argument("-e", "--end", type=int, required=True, help="index of last image to do matching") Thank you very much ! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.
hurricane2018
Hi, thank you for your wonderful work and code. However, when I run the "python match.py", the error occurs. Could you give more details about the parameters including data_dir, save_dir, t, s, and e?
======== parser.add_argument("--data_dir", type=str, required=True, help="path to root dir to data.") parser.add_argument("--save_dir", type=str, required=True, help="path to root dir to save results") parser.add_argument("-t", "--tag", type=str, required=True, help="tag used to indicate one run") parser.add_argument("-s", "--start", type=int, required=True, help="index of first image to do matching,\ this is used for parallel matching of different images") parser.add_argument("-e", "--end", type=int, required=True, help="index of last image to do matching")
Thank you very much !