Primary competition visual

Lacuna Malaria Detection Challenge

Helping Uganda
$5 000 USD
Completed (over 1 year ago)
Object Detection
Computer Vision
1331 joined
368 active
Starti
Aug 16, 24
Closei
Nov 17, 24
Reveali
Nov 17, 24
Are we obliged to use yolov8?
Notebooks · 1 Nov 2024, 03:03 · 6

Hello,

Please, Are we obliged to use Yolov8 as it's in the starter notebook?

I am new to Yolo and this class of image classification problems but I tried most of the things I know (except changing the architecture of Yolov8 because I don't know how to do that) but I still have around of 40% on the test data.

Are we obliged to use YoloV8?

Discussion 6 answers

I don't think so. I believe any open source resources can be used.

1 Nov 2024, 06:51
Upvotes 3
User avatar
MuhammadQasimShabbeer
Engmatix

yes you can use any model which can perform on this tasks

1 Nov 2024, 07:57
Upvotes 1

I think it may give the best performance, ensembling others may be worth an effort

1 Nov 2024, 10:42
Upvotes 2
User avatar
MuhammadQasimShabbeer
Engmatix

You can Try one of these Latest Yolov version is 11 you can set it in the code Here https://www.kaggle.com/code/muhammadqasimshabbir/zindi1-2ghana-crop-disease from my note book you can set it before setting the training parameter in like this # Load a YOLO pretrained model

model = YOLO('yolo11m.pt')

YOLO Versions 8 to 11

  1. YOLOv8Sizes Available: Small (YOLOv8s), Medium (YOLOv8m), Large (YOLOv8l), Extra Large (YOLOv8x). Description: Designed for high performance and accuracy, YOLOv8 introduces improved architectures that enhance detection capabilities across various environments.
  2. YOLOv9Sizes Available: Small, Medium, Large, Extra Large. Description: Builds on the advancements of YOLOv8 with further optimizations, offering better speed and efficiency for real-time object detection tasks.
  3. YOLOv10Sizes Available: Small, Medium, Large, Extra Large. Description: Focuses on refining detection accuracy while maintaining speed, making it suitable for both edge devices and powerful servers.
  4. YOLOv11Sizes Available: Small, Medium, Large, Extra Large. Description: Represents the latest advancements in YOLO technology, emphasizing versatility and adaptability in various detection scenarios, from small objects to complex environments.
User avatar
MuhammadQasimShabbeer
Engmatix

Your are welcome