Track

Computer Vision

Goal: Image classification, detection, and segmentation with convolutional networks and vision transformers.

Prereqs: Deep Learning.

Status: done

Work through the steps in order. Bold links open YouTube.

Step	Concept	YouTube	Read
1	Convolution	3Blue1Brown — But what is a convolution?	d2l — From fully connected layers to convolutions
2	CNN image classification	StatQuest — Image Classification with Convolutional Neural Networks	cs231n — Convolutional neural networks
3	Classic CNN architectures	Stanford CS231n — Convolutional Neural Networks	Krizhevsky et al. 2012 — AlexNet (NeurIPS PDF)
4	Residual networks	Insightforge — How ResNet Solved the Vanishing Gradient	He et al. 2015 — Deep Residual Learning
5	Transfer learning	Aladdin Persson — PyTorch Transfer Learning and Fine Tuning	cs231n — Transfer learning
6	Image augmentation		d2l — Image augmentation
7	Object detection (two-stage)	Stanford CS231n — Detection and Segmentation	Ren et al. 2015 — Faster R-CNN
8	YOLO (one-stage detection)	DeepBean — How YOLO Object Detection Works	Redmon et al. 2015 — You Only Look Once
9	Semantic segmentation		Long et al. 2014 — Fully Convolutional Networks
10	Instance segmentation		He et al. 2017 — Mask R-CNN
11	Vision transformers	Yannic Kilcher — An Image is Worth 16x16 Words	Dosovitskiy et al. 2020 — Vision Transformer
12	Fine-tuning practice		d2l — Fine-tuning

Found a broken link or an unclear step? Report a problem with this track.