Ant Datasets for Oso-Hormiguero

This repository contains a collection of datasets related to ants used in the Oso-Hormiguero project, along with scripts to process the data for use with the selected model.

Contents

• Generators: Scripts to generate different combinations of datasets mentioned in the training log. Located under generators folder.
• Processors: Scripts to normalize the data for a specific model, named <dataset>_<model>.py. Currently only works for yolo.

How to Use

1. Run python main.py to start the CLI client for processing datasets and/or generating mixes.
2. Select one of the options prompted in the console:
   • Option 1: Process a dataset for YOLO:
     • Normalizes the datasets for YOLOv8.
   • Option 2: Create a mix of data sets.
     • Combines all the project datasets into one.

Option 1 - Process a Dataset for YOLO

• The console will display the available datasets. Select the datasets you want to mix, separating them by semicolons.

Option 2 - Create Mixes

• Select one of the options prompted in the console:
  • Mix all the datasets: Uses the entire dataset mix.
  • Mix all the datasets with less slope: Reduces the number of similar images to decrease the slope.
  • Custom mix: Displays the available datasets. Select the datasets you want to mix, separating them by semicolons.

Results

Under reults folder there are two directorys:

datasets

Contains the processed datasets.

mixes

Contains the mixed datasets.

Datasets References

• Cao, Xiaoyan (2021), “ANTS--Ant Detection and Tracking”, Mendeley Data, V3, doi: 10.17632/9ws98g4npw.3, visited on 2024-03-24
• Djay de Gier, Ant Object Detection Dataset, Roboflow, visited on 2024-03-24