Image Synthesis Pipeline

---

This pipeline includes all the needed tools for synthesizing images using the cut-and-paste approach in overlapping wheat spikes on the background images.

Code configuration and run:

To implement the code we used Python version 3.9, and a version of the packages can be installed through installing all the required packages.

1. In order to install the required packages please run the following command using pip:

      pip install -r requirements.txt

In order to run each python script you first need to change the related configuration file in the config/ folder for each file.

2. In order to extract frames from videos:
   1. Change the configuration file named frames_extractor.yaml in the configs/ folder.
   2. Run the following command in the terminal:

      python3 frames_extractor.py --config configs/frames_extractor.yaml

3. In order to extract real and fake objects from the segmented representative image:
   1. Change the configuration file named objects_extractor.yaml in the configs/ folder.
   2. Run the following command in the terminal:

      python3 objects_extractor.py --config configs/objects_extractor.yaml

4. In order to simulate the dataset using the previously extracted background frames and real and fake objects:
   1. Change the configuration file named simulation.yaml in the configs/ folder.
   2. Run the following command in the terminal:

      python3 simulation.py --config configs/simulation.yaml

📝 Cite

@article{
    doi:10.34133/plantphenomics.0025,
    author = {Keyhan Najafian  and Alireza Ghanbari  and Mahdi Sabet Kish  and Mark Eramian  and Gholam Hassan Shirdel  and Ian Stavness  and Lingling Jin  and Farhad Maleki },
    title = {Semi-Self-Supervised Learning for Semantic Segmentation in Images with Dense Patterns},
    journal = {Plant Phenomics},
    volume = {5},
    number = {},
    pages = {0025},
    year = {2023},
    doi = {10.34133/plantphenomics.0025},
    URL = {https://spj.science.org/doi/abs/10.34133/plantphenomics.0025},
    eprint = {https://spj.science.org/doi/pdf/10.34133/plantphenomics.0025}
}