Hello, Thank you very much for your Git

[huang229]

What is the data storage format of the label TXT file？For example, there are multiple targets in an image, and how it corresponds to the label TXT file. I can't open the downloaded data and show the damage.Thank you.

[liulei01]

In folder "train_data", each image file has a corresponding .rbox file. For example, an image file 1.tif would be labeled by 1.tif.rbox. Supposing that there are two targets in this image, the rbox file should have two lines and each line corresponds to one target. For example, it may looks like:

128.0 117.0 50.0 60.0 1 16.5 262.0 152.0 87.0 90.0 1 250.0

The 5 numbers in each line are center_x, center_y, width, height, label, angle, respectively. To build a LMDB database, you should use a list file to indicate the relationship between each image file and its rbox file. For example, in trainval.txt, you should write:

1.tif 1.tif.rbox 2.tif 2.tif.rbox

if you have two images for training.

[liulei01]

The downloaded data should be extracted using command cat data.tar.gz*>data.tar.gz first to merge each subfile together.

[huang229]

Thank you very much for your help.I wish you a good mood every day.

[jeejeelee]

@liulei01 your dataset is so big for me, can you upload smaller dataset( eg. about 2G ),I just want to learn the process of training with your dataset thank you ~~~

[liulei01]

@jeejeeli Do you have difficulty downloading them or training with all of them? If the latter, you can simply delete lines in trainval.txt to decrease the number of training samples.

[jeejeelee]

@liulei01 thanks for your answer， it is difficult for me to download your data, but now I have downloaded your dataset using VIP ,hhh~

[pfchenstuxidianedncn]

Have you download the datasets and extracted them? I have some problem when I extracting them.@huang229

[huang229]

@pfchenstuxidianedncn ubuntu system command: cat data.tar.gz*>data.tar.gz

[pfchenstuxidianedncn]

Yes, I have already carried out this command, but I have problem when I extract the "data.tar.gz". How can I solve this problem.

[cyrilli]

Hi, I managed to compile and train your model on Airplane data, and I am about to build my own dataset. I am wondering what is BAIYUNJICHANG_Level_19_0.tif.pos.txt in DRBox/data/Airplane/train_data. Is this file needed to create a dataset?