A lot of wonderful datasets are now available online, such as COCO or Imagenet. which is challenging the limits of computer vision. But it's not easy for us to do some small experiments with such a large number of images to quickly test the validity of algorithmn. For this reason, I created a small dataset named "yymnist" to do both classification and object detection.
classification | object detection |
---|---|
Clone this repo and install some dependencies
$ pip install opencv-python
$ git clone https://github.com/YunYang1994/yymnist.git
If you want to use this dataset for classification, just use mnist
folder. Each image in the folder starts with a name of its label. For example:
mnist/train/0_32970.pgm -> 0
mnist/train/5_12156.pgm -> 5
$ python yymnist/make_data.py
$ python yymnist/show_image.py # [option]
you will defaultly get 1000 pictures in the folder ./yymnist/Images
and a label.txt
in the ./yymnist/
. As for label.txt
# image_path x_min, y_min, x_max, y_max, class_id x_min, y_min ,..., class_id
此处可能存在不合适展示的内容,页面不予展示。您可通过相关编辑功能自查并修改。
如您确认内容无涉及 不当用语 / 纯广告导流 / 暴力 / 低俗色情 / 侵权 / 盗版 / 虚假 / 无价值内容或违法国家有关法律法规的内容,可点击提交进行申诉,我们将尽快为您处理。