Главная
Study mode:
on
1
- Introduction
2
- Overview of DR and how to detect
3
- A look at the data
4
- Creating a baseline solution
5
- Result from baseline
6
- Idea #1: Preprocessing
7
- Result #1
8
- Idea #2: Loss Function
9
- Result #2
10
- Idea #3: Balanced Loader skipped
11
- Idea #4: Augmentation
12
- Result #4
13
- Idea #5: Using left and right information
14
- Result #5
15
- Idea #6: Increase image resolution
16
- Result #6 from various resolutions
17
- Final Result and Ending
Description:
Explore a comprehensive tutorial on detecting eye disease using deep learning, focusing on a top 1% Kaggle solution without ensemble methods. Learn about diabetic retinopathy detection, data analysis, and the step-by-step process of building a high-performing model. Discover various techniques including preprocessing, loss function optimization, data augmentation, and leveraging left and right eye information. Follow along as the instructor demonstrates how to create a baseline solution and iteratively improve it through multiple ideas and experiments. Gain insights into the impact of image resolution on model performance and understand the final results achieved using these advanced deep learning techniques.

Detecting Eye Disease Using Deep Learning - Kaggle Top 1% Solution, No Ensemble

Aladdin Persson
Add to list