dasmehdix/drone-dataset

Drone dataset to guide enemy drones (with some tools)

drone-dataset

Drone dataset to lock on enemy drones

This dataset prepared for academical and competition purpose. Main idea of this dataset is to guide our drone to other drones. Dataset includes ~1400 drone images and label files. The label files are in format of both .txt and .xml. So, You can use it with Tensorflow, Darknet and PyTorch too. Most of images taken from Google and Yandex's image search toolbox with image(web) scrapers. The other images are cropped from drone videos that downloaded from Youtube. Results are pretty good(%86mAP).

Here is the link of dataset

Is it working really?!

Answer: Yes, even with tiny one like above!

Some Results

500 Photo Yolo-v3 weights

Full Dataset Darknet53 weights

Full Dataset Yolo-v3 weights (tuned)

Full Dataset Yolov3-tiny weights

Full Dataset Yolov3-tiny weights(tuned)

Bonus-1

The xmlparser is written to solve a issue that occured in labelimg. When I labeled photos with LabelImg, there was no problem firstly. Then, when I changed the path of images(and format of images like jpg or png) the xml files become a problem. Because the path in the label xml file and the image path looks diffrent. So, instead of labelling the photos again, kubilay write a script to change the path in the label xml files, so easyly we can train our images in every PC.

Bonus-2

I also added a python script(txt_path_generator_yolo) to write the path of images in a txt file to train YOLO-v3.

Suggestion

Images are in diffrent format like jpg, png or jpeg. I suggest you to convert all of them to jpg format. For windows,

1. Open cmd
2. Go to the path that contain images
3. Run this command:

ren *. *.jpg

You can cite me from google scholar or researchgate. Or, just mention link of the dataset which is enough for me. This dataset is freely available for academical and fun purpose :)

drone-dataset's Issues

problems while training with dataset

Hey
Thanks for your dataset.
Unfortunately I have some issues..
I am working with tensorflow and took your Images as dataset.

When I want to try the model I get the following error:

flu@flubuntu:~/Tensorflow/scripts/preprocessing$ python3 generate_tfrecord.py --label=drone --csv_input=/home/flu/Tensorflow/workspace/training_demo/annotations/test_labels.csv --img_path=/home/flu/Tensorflow/workspace/training_demo/images/test --output_path=/home/flu/Tensorflow/workspace/training_demo/annotations/test.record
/home/flu/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:523: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
  _np_qint8 = np.dtype([("qint8", np.int8, 1)])
/home/flu/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:524: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
  _np_quint8 = np.dtype([("quint8", np.uint8, 1)])
/home/flu/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:525: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
  _np_qint16 = np.dtype([("qint16", np.int16, 1)])
/home/flu/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:526: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
  _np_quint16 = np.dtype([("quint16", np.uint16, 1)])
/home/flu/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:527: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
  _np_qint32 = np.dtype([("qint32", np.int32, 1)])
/home/flu/.local/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:532: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'.
  np_resource = np.dtype([("resource", np.ubyte, 1)])
Traceback (most recent call last):
  File "generate_tfrecord.py", line 117, in <module>
    tf.app.run()
  File "/home/flu/.local/lib/python3.6/site-packages/tensorflow/python/platform/app.py", line 125, in run
    _sys.exit(main(argv))
  File "generate_tfrecord.py", line 108, in main
    tf_example = create_tf_example(group, path)
  File "generate_tfrecord.py", line 64, in create_tf_example
    image = Image.open(encoded_jpg_io)
  File "/usr/lib/python3/dist-packages/PIL/Image.py", line 2590, in open
    % (filename if filename else fp))
OSError: cannot identify image file <_io.BytesIO object at 0x7fb7f8d869e8>

Can you help me with that?

Thx so much

A question about Dataset

Hi there,
Thanks for the dataset. I have downloaded dataset from https://www.kaggle.com/dasmehdixtr/drone-dataset-uav

There are two sets, one is xml format and other is txt format. My question is: Are the pictures are same in both sets or different?