Official ASF-YOLO

This is the source code for the paper "ASF-YOLO: A Novel YOLO Model with Attentional Scale Sequence Fusion for Cell Instance Segmentation" published on Image and Vision Computing (IMAVIS), of which I am the first author. The paper is available to download on ScienceDirect or arXiv.

Model

The Attentional Scale sequence Fusion You Only Look Once (ASF-YOLO) model configuration (i.e., network construction) file is asf-yolo.yaml in the directory ./models/segment.

Training

The hyperparameter setting file is hyp.scratch-low.yaml in the directory ./data/hyps/.

Installation

Install requirements.txt in a Python>=3.8.0 environment, including PyTorch>=1.8.

pip install -r requirements.txt

Training CLI

python segment/train.py

Testing CLI

python segment/predict.py

Evaluation

We trained and evaluated ASF-YOLO on the two datasets: the 2018 Data Science Bowl (DSB2018) from Kaggle and the Breast Cancer Cell (BCC) dataset from the Center for Bio-Image Informatics, University of California, Santa Barbara (UCSB CBI). The yolov5l-seg.pt is the initial weight of the pretrained MS COCO dataset by YOLOv5l and isn't the ASF-YOLO training weight on the evaluation datasets, which has been stated in the paper.

Referencing Guide

Please cite our paper if you use code from this repository. Here is a guide to referencing this work in various styles for formatting your references:

Plain Text

• Elsevier Numbered Style
  M. Kang, C.-M. Ting, F.F. Ting, R.C.-W. Phan, ASF-YOLO: A novel YOLO model with attentional scale sequence fusion for cell instance segmentation, Image Vis. Comput. 147 (2024) 105057.
  

• Harvard (Name–Date) Style
  Kang, M., Ting, C.-M., Ting, F.F., Phan, R.C.-W., 2024. ASF-YOLO: A novel YOLO model with attentional scale sequence fusion for cell instance segmentation. Image Vis. Comput. 147, 105057.
  

• IEEE Reference Style
  M. Kang, C.-M. Ting, F. F. Ting, and R. C.-W. Phan, "Asf-yolo: A novel yolo model with attentional scale sequence fusion for cell instance segmentation," Image Vis. Comput., vol. 147, 105057, Jul. 2024.
  

• Nature Referencing Style
  Kang, M., Ting, C.-M., Ting, F. F. & Phan, R. C.-W. ASF-YOLO: a novel YOLO model with attentional scale sequence fusion for cell instance segmentation. Image Vis. Comput. 147 105057 (2024).
  

• Springer Reference Style
  Kang, M., Ting, C.-M., Ting, F.F., Phan, R.C.-W.: ASF-YOLO: a novel YOLO model with attentional scale sequence fusion for cell instance segmentation. Image Vis. Comput. 147, 105057 (2024)
  

BibTeX Format

\begin{thebibliography}{1}
\bibitem{1} M. Kang, C.-M. Ting, F.F. Ting, R.C.-W. Phan, ASF-YOLO: A novel YOLO model with attentional scale sequence fusion for cell instance segmentation, Image Vis. Comput. 147 (2024) 105057.
\end{thebibliography}

\begin{thebibliography}{1}
\bibitem[Kang(2024)]{Kang24} Kang, M., Ting, C.-M., Ting, F.F., Phan, R.C.-W., 2024. ASF-YOLO: A novel YOLO model with attentional scale sequence fusion for cell instance segmentation. Image Vis. Comput. 147, 105057.
\end{thebibliography}

@article{Kang24Asfyolo,
  author = "Ming Kang and Chee-Ming Ting and Fung Fung Ting and Rapha{\"e}l C.-W. Phan",
  title = "ASF-YOLO: A novel yolo model with attentional scale sequence fusion for cell instance segmentation",
  journal = "Image Vis. Comput.",
  volume = "147",
  % ieee_fullname.bst
  pages = "105057",
  % IEEEbib.bst
  note = "p. 105057", 
  month = "Jul.",
  year = "2024",
}

@article{Kang24Asfyolo,
  author = "Kang, Ming and Ting, Chee-Ming and Ting, Fung Fung and Phan, Rapha{\"e}l C.-W.",
  title = "ASF-YOLO: a novel YOLO model with attentional scale sequence fusion for cell instance segmentation",
  journal = "Image Vis. Comput.",
  volume = "147",
  pages = "105057",
  publisher = "Elsevier",
  address = "Amsterdam",
  year = "2024",
  doi= "10.1016/j.imavis.2024.105057",
  url = "https://doi.org/10.1016/j.imavis.2024.105057"
}

^{NOTE: Please remove some optional BibTeX fields, for example, series, volume, address, url and so on, while the LaTeX compiler produces an error. Author names may be manually modified if not automatically abbreviated by the compiler under the control of the .bst file if applicable which defines bibliography/reference style. Kang24Asfyolo could be b1, bib1, or ref1 when references appear in numbered style in which they are cited. The quotation mark pair "" in the field could be replaced by the brace {}.}

License

ASF-YOLO is released under the GNU Affero General Public License v3.0 (AGPL-3.0). Please see the LICENSE file for more information.

Copyright Notice

Many utility codes of our project base on the codes of Ultralytics YOLOv5, EIoU and Soft-NMS repositories.