Skip to content

identify polyps in colonoscopy footage using YOLOv8x (convolutional neural network)

Notifications You must be signed in to change notification settings

moonsujo/transfer_learning_yolo

Repository files navigation

Transfer Learning and Computer Vision

Fine tune a pre-trained computer vision model to test the effectiveness of transfer learning.

Domain: medical field - colonoscopy. 27k training images, 4k validation

Model: YOLOv8x

In order to obtain a higher accuracy, I went a step further and performed hyperparameter variation experiments.

Batch size: 8, 16

Learning rate pairs (initial, final): (0.001, 0.000001), (0.0001, 0.01)

Highest accuracy obtained (mAP): 95.8%

Best performing variation: batch size of 16, learning rate pair of (0.0001, 0.01)

About

identify polyps in colonoscopy footage using YOLOv8x (convolutional neural network)

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages