[Paper]
- Simple to train architecture for efficient malaria detection
- STATE-OF-THE-ART for parasite detection in [
MP-IDB,IML,M5]
This table compares the detection performance of our proposed framework against the state-of-the-art on different datasets.
| Dataset | Species | Work | Reference Model | AP | Increase AP% |
|---|---|---|---|---|---|
| M5 | P. Falciparum | Sultani et al. (2022)~[1] | Faster R-CNN | 66.8 | - |
| M5 | P. Falciparum | Proposed Framework | YOLOv5-SPAM-3H | 71.0 | 4.2 |
| MP-IDB | P. Falciparum | Zedda et al.~[2] | YOLOv5m6 | 62.5 | - |
| MP-IDB | P. Falciparum | Proposed Framework | YOLOv5-SPAM-AH | 86.5 | 23.0 |
| MP-IDB | P. Malariae | Zedda et al.~[2] | YOLOv5m6 | 80.0 | - |
| MP-IDB | P. Malariae | Proposed Framework | YOLOv5-SPAM-AH | 94.9 | 14.9 |
| MP-IDB | P. Ovale | Zedda et al.~[2] | YOLOv5m6 | 83.9 | - |
| MP-IDB | P. Ovale | Proposed Framework | YOLOv5-SPAM-MH | 95.1 | 11.2 |
| MP-IDB | P. Vivax | Zedda et al.~[2] | YOLOv5m6 | 83.1 | - |
| MP-IDB | P. Vivax | Proposed Framework | YOLOv5-SPAM-AH | 88.3 | 5.2 |
| IML | P. Vivax | Proposed Framework | YOLOv5-SPAM-3H | 67.4 | - |
Before using the code, make sure to follow these setup instructions:
Extract the "ultralytics" zip file and place it in your Python packages folder.
The full list of model configurations can be found in the "config" folder.
The full list of data configurations can be found in the "data" folder.
A small usage example is provided in the train_notebook.ipynb notebook.
Feel free to contribute by adding more papers, improving code, or providing feedback. Open issues and pull requests are welcome!
This project is licensed under the MIT License.
[1] Sultani et al. (2022), Link to Paper M5
[2] Zedda et al. (ICIAP MALARIA), Link to Paper ICIAP
If you use YOLO-PAM in your research or wish to refer to the baseline results published in the original paper, please use the following BibTeX entry.
@article{zedda_yolo-pam_2023,
title = {{YOLO}-{PAM}: {Parasite}-{Attention}-{Based} {Model} for {Efficient} {Malaria} {Detection}},
volume = {9},
copyright = {http://creativecommons.org/licenses/by/3.0/},
issn = {2313-433X},
shorttitle = {{YOLO}-{PAM}},
url = {https://www.mdpi.com/2313-433X/9/12/266},
doi = {10.3390/jimaging9120266},
language = {en},
number = {12},
urldate = {2023-11-30},
journal = {Journal of Imaging},
author = {Zedda, Luca and Loddo, Andrea and Di Ruberto, Cecilia},
month = dec,
year = {2023},
note = {Number: 12
Publisher: Multidisciplinary Digital Publishing Institute},
keywords = {computer vision, deep learning, early malaria diagnosis, image processing, malaria parasite detection},
pages = {266},
}