Detecting Defects in Steel Sheets with Computer Vision

What you'll be able to do

• How to solve deep learning industrial project end to end?
• What is computer vision?
• Understand the U-Net model architecture.
• How to choose right evaluation metric?
• What is BCE Logits Loss and code implementation?
• What is Dice Loss and code implementation?
• How to visualize with Tensorboard?

1. 01Introduction3 lessons · 13 min

   1. Defect Detection in ManufacturingPreview4:44
   2. Download Resources1:00
   3. Data OverviewPreview6:57
2. 02U-Net Model Architecture2 lessons · 21 min

   1. Encoder: Going Down the U-NetPreview15:57
   2. Decoder: Moving Up the U-NetPreview4:36
3. 03Evaluation Metrics3 lessons · 8 min

   1. What is Dice?Preview3:18
   2. Cross Entropy vs. DicePreview3:28
   3. Course Review1:00
4. 04Setup, Helper Functions and Explore Data (Code Demo)5 lessons · 19 min

   1. Setup Environment and Import Data3:29
   2. Helper Functions for Encoding and Decoding Masks with OpenCV12:30
   3. Helper Function for Dice Loss Calculation1:12
   4. Set Device Configuration0:20
   5. Exploring Data1:58
5. 05Train Model with BCE Logits Loss (Code Demo)7 lessons · 15 min

   1. Define Hyperparameters0:52
   2. Initialize Dataloader1:08
   3. Initialize UNet Model7:11
   4. Define Loss and Optimizer0:25
   5. Train Model1:50
   6. Evaluate Model1:29
   7. Check performance when training with entire data2:07
6. 06Train Model with BCE Logits Loss and Learning Rate Scheduler (Code Demo)1 lesson · 5 min

   1. Train Model with BCE Logits Loss and Learning Rate Scheduler (Code Demo)5:14
7. 07Train Model with Dice Loss and Learning Rate Scheduler (Code Demo)1 lesson · 2 min

   1. Train Model with Dice Loss and Learning Rate Scheduler (Code Demo)1:49
8. 08Visualize with TensorBoard (Code Demo)4 lessons · 5 min

   1. Initial Setup0:28
   2. Initialize Writer0:19
   3. Show Images in TensorBoard2:01
   4. Show Model Architecture in TensorBoard2:21

What you need

• Basics of Python
• Foundational knowledge of Data Science
• High school maths

Who this course is for

• Data Science Aspirants
• Data Science Professionals
• Software/Data engineers interested in quantitative analysis
• Professionals working with large datasets
• Data analysts, economists, researchers

Your instructor

Prabhakaran S

Principal Data Scientist

My name is Selva, and I am super excited to mentor you on this course!

I head the Data Science team for a global Fortune 500 company and over the last 10 years of my data science experience I've deployed 20+ global products. I'm also the Founder and Chief Author of Machine Learning Plus, which has over 4M annual readers.

I specialize in covering the in-depth intuition and maths of any concept or algorithm. And based on my existing student requests, I've put up the series of courses and projects with detailed explanations, just like on-the-job experience. Hope you love it!

from 3,965 ratings

4.92

learners

57K+

courses

85