How to use the dataset to train a YOLO object detector? about pie HOT 5 CLOSED

Hi @Christoph9402,

1. You can try loading the annotations into (CVAT)[https://github.com/openvinotoolkit/cvat] and exporting them in YOLO format.
2. Heading direction and speed are in annotations_vehicle.zip, which contains one xml per video with direction, current speed and more. A simple Python script to parse those and output them to csv should do it or you can load each xml file into Excel (or LibreOffice Calc) (see (link)[https://www.spreadsheetweb.com/how-to-import-xml-files-into-excel/]).

from pie.

Thank you for your answer! I am going to try the tool for converting tomorrow and keep you updated. Your Idea to load the xml files into Excel was just what i was looking for :)

from pie.

I have an additional question regarding the dataset. I downloaded the Clips and Annotations and have the following folder structure:

PIE_dataset
    annotations
        set01
        set02
        ...
    PIE_clips
        set01
        set02
        ...

I now place the files pie_data.py , split_clips_to_frames.sh and a python file extract_images.py which includes these lines of code, since i only want to save images with bounding boxes

from pie_data import PIE
pie_path = 'C:/Users/Chris/Documents/Uni/Studienarbeit/Dataset/PIE_dataset'
imdb = PIE(data_path=pie_path)
imdb.extract_and_save_images(extract_frame_type='annotated')

inside the PIE_dataset folder. When I run extract_images.py, I get the following error message:

from scipy import optimize
  File "C:\Users\Chris\anaconda3\envs\studienarbeit_env\lib\site-packages\scipy\optimize\__init__.py", line 398, in <module>
    from ._nnls import nnls
ImportError: DLL load failed: The specified module could not be found

What did I do wrong? How can I fix this? Also, is the folder structure wrong for running the script?

from pie.

The folder structure seems ok. Based on the error message, something may be wrong with your scipy installation. I don't use conda or Windows so I don't know how to resolve it. Try googling the error message or check issues in conda and scipy githubs.

from pie.

@ykotseruba Thank you very much for your help :) I am going to find an answer to that on google

from pie.