A Friendly Introduction to Deep Learning for Computer Vision

on YouTube ↗59m

" summary

Starts from perceptron math and shows exact code using PyTorch tensors, datasets and dataloaders, a linear model with mean squared error and SGD (epochs ~20) to learn a 3×3 “nine square” rule, scales to MNIST with a CNN reaching ~97% accuracy, then uses transfer learning to train a taco vs burrito classifier, exports the model to ONNX and deploys it to AKS where the endpoint returns predictions (example: burrito ~0.73).

AI-written summary · about

" chapters (14)

1. 0:00Start
2. 1:17Agenda
3. 2:14Machine Learning
4. 2:38What Machine Learning Does
5. 5:09The Nine Square Problem
6. 6:09Nine Square Problem
7. 8:23Tensors Sets and Data Loaders
8. 8:34Jupiter Notebooks
9. 16:11Creating a Model
10. 23:18Linear Model
11. 35:48Mean Squared Error
12. 43:08Padding
13. 46:02Convolutional Neural Network
14. 57:18Azure Machine Learning

" related talks

• 01h 00mA gentle introduction to Deep LearningLaurent Bugnion · NDC Porto 2022
• 01h 05mDeep Learning with PyTorchSeth Juarez · NDC London 2019
• 01h 29mComputer Vision: An (Un?)Expected Journey, with Keras and TensorflowRodolfo Bonnin · PyData London 2018
• 01h 04mWhat you always wanted to know about Deep Learning, but were afraid to askWei-Meng Lee · NDC London 2020
• 01h 27mHands-On Computer Vision with PyTorchRobert Alvarez · PyData NYC 2022
• 01h 26mDeep Neural Networks with PyTorchStefan Otte · PyData Berlin 2018
• 47mDeep Learning: a practical introduction from an Android dev perspectiveJoaquim Verges · droidcon SF 2019
• 54mIntroduction to Understanding Deep LearningAntony Ross · Silicon Valley Code Camp 2018