mkang315 / bgf-yolo Goto Github PK

Official implementation of "BGF-YOLO: Enhanced YOLOv8 with Multiscale Attentional Feature Fusion for Brain Tumor Detection".

License: GNU Affero General Public License v3.0

Python 99.47% Shell 0.53%

afpn bi-level-routing-attention biformer bifpn br35h-brain-tumor-detection-2020 cbam ciou-loss computer-vision-algorithms deep-learning-framework deep-neural-networks gfpn giou-loss lesion-detection medical-image-analysis medical-image-computing medical-image-detection medical-image-processing object-detection yolo wiou-loss

bgf-yolo's Introduction

Official BGF-YOLO

This is the source code for the paper, "BGF-YOLO: Enhanced YOLOv8 with Multiscale Attentional Feature Fusion for Brain Tumor Detection", of which I am the first author.

Model

The model configuration (i.e., network construction) file is bgf-yolo.yaml in the directory ./models/bgf.

The hyperparameter setting file is default.yaml in the directory ./yolo/cfg/.

Installation

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

pip install -r requirements.txt  # install

Training CLI

python yolo/bgf/detect/train.py

Testing CLI

python yolo/bgf/detect/predict.py

Evaluation

We trained and evaluated BGF-YOLO on the dataset Br35H :: Brain Tumor Detection 2020. The .txt format annotations in the file dataset-Br35H.zip are coverted from original json format.

Model	Precision	Recall	mAP₅₀	mAP_50:95
RT-DETR-X	0.825	0.770	0.870	0.597
Co-DETR with Swin L (36 Epochs, DETR Augmentation)	–	–	0.941	0.609
YOLOv9-E	0.927	0.869	0.919	0.630
BGF-YOLO (Ours)	0.919	0.926	0.974	0.653

Generalizability in External Validation

We conducted additional experimental validation on a different domain using the COVID-19 facemask detection dataset. The table below consistently shows the superior detection performance of our method compared to YOLOv8x. This indicates the generalizability of our method to other domains of object detection.

Model	Precision	Recall	mAP₅₀	mAP_50:95
YOLOv8x	0.866	0.773	0.802	0.494
BGF-YOLO (Ours)	0.847	0.764	0.820	0.504

License

BGF-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 YOLOv8, GiraffeDet, DAMO-YOLO, and BiFormer repositories.

bgf-yolo's People

Contributors

Stargazers

Watchers

Forkers

theteaterra bafoxw huihui308 lidexf cv-det bushrabazai shiaoyoungcui zz4fap ankita241103

bgf-yolo's Issues

Training report error

1.root@7s26acs0ob5sl-0:/E22201117/clk/code/BGF-YOLO-main# python yolo/bgf/detect/train.py
Traceback (most recent call last):
File "yolo/bgf/detect/train.py", line 7, in
from ...nn.tasks import DetectionModel
ValueError: attempted relative import beyond top-level package
2.BGF-YOLO/yolo/init.py
from . import bfg
all = 'bfg', # tuple or list

May I ask if it should be bgf?
from . import bgf
all = 'bgf', # tuple or list

parameters of model

Hi,
I've looked at this perfect model for detecting the brain tumors and the results was very intresting for me. To be honest, I'm looking for the number parameters of this model. Can you please help me and make it clear for me that how many parameters does your model consist?

dataset

Is it possible to release segmented datasets, similar to what RCS-YOLO has disclosed?

When running "python yolo/bgf/detect/train.py", I encountered the above problem. How to solve it?

Traceback (most recent call last):
File "yolo/bgf/detect/train.py", line 7, in
from ...nn.tasks import DetectionModel
ImportError: attempted relative import with no known parent package

Request for pretrained weight

Good morning,

Would you mind sharing the pretrained weight of Br35h of YOLO-V8 and BGF-YOLO, please?
Thank you very much in advanced.

problem with relative imports and missing files

I try to run the train script:

❯ python yolo/bgf/detect/train.py

And get this error:

Traceback (most recent call last):
  File "/ssdpool/projects//BGF-YOLO/yolo/bgf/detect/train.py", line 7, in <module>
    from ...nn.tasks import DetectionModel
ImportError: attempted relative import with no known parent package

I tried replacing all relative imports with absolute imports and also setting the PYTHONPATH to the project dir or running the script as a module. However I run into more and more problems. E.g. this one:

❯ python -m yolo.bgf.detect.train

Traceback (most recent call last): 
File "/home/thomas/conda/envs/timm/lib/python3.10/runpy.py", line 187, in _run_module_as_main
mod_name, mod_spec, code = _get_module_details(mod_name, _Error)
  File "/home/thomas/conda/envs/timm/lib/python3.10/runpy.py", line 110, in _get_module_details
    __import__(pkg_name)
  File "/ssdpool/projects/BGF-YOLO/yolo/__init__.py", line 8, in <module>
  from yolo.engine.model import YOLO                                                                                                                                                                                                                                              File "/ssdpool/projects/BGF-YOLO/yolo/engine/model.py", line 21, in <module>
    ClassificationModel, yolo.v8.classify.ClassificationTrainer, yolo.v8.classify.ClassificationValidator,
    AttributeError: partially initialized module 'yolo' has no attribute 'v8' (most likely due to a circular import)

If I search the source for ClassificationTrainer it's nowhere to be found. I think you may have missed some parts of the source.
They would likely be available in the original yolov8 source, however since this is probably on a different version I was hesitant to grab it from there.

I already follow your notes. But I just wonder how to running it.

In yolov5 to use other loss function, we need to change CIOU=False.

Thank you