Skip to main content

Computer Vision Track

Computer Vision Track

Goal​

Focus: Teaching machines to see and recognize.

Curriculum​

Core Computer Vision & Deep Learning​

Topics:

  • Image representation (pixels, channels, tensors)
  • Convolutional Neural Networks (CNNs)
  • Filters, kernels, padding, stride
  • Feature extraction and hierarchical representations
  • Activation functions (ReLU, sigmoid, softmax)
  • Loss functions for vision tasks
  • Backpropagation in CNNs
  • Optimization methods (SGD, Momentum, Adam)
  • Regularization (dropout, data augmentation, batch normalization)
  • Transfer learning and fine-tuning
course
advanced

CS231n: Convolutional Neural Networks for Visual Recognition (Winter 2016) Lectures 4–7

8-10 hours

Hands-On Deep Learning with PyTorch​

Topics:

  • PyTorch tensor operations and automatic differentiation
  • Building neural networks using nn.Module
  • Forward and backward passes in practice
  • Loss functions and optimization loops
  • Training and evaluation workflows
  • GPU acceleration and device management
  • Debugging and inspecting training behavior
  • Saving, loading, and reusing trained models
tutorial
intermediate

UvA Deep Learning 1 (PyTorch) – Tutorials 3–6

6-8 hours

Object Detection Models & Architectures​

Topics:

  • Object detection vs image classification
  • Bounding boxes and localization
  • Intersection over Union (IoU)
  • Anchor boxes and grid-based detection
  • Region proposal methods
  • One-stage vs two-stage detectors
  • Detection loss functions
  • Evaluation metrics (mAP, precision, recall)
course
advanced

CS231n Lectures (8~14)

6-8 hours

Single-Stage Detectors (YOLO Family)​

Topics:

  • YOLO problem formulation
  • Grid-based prediction
  • Anchor boxes
  • Bounding box regression
  • YOLO loss function
  • Real-time detection constraints
paper
advanced

YOLOv1 & YOLOv2

6-8 hours

Practical Object Detection (YOLOv11)​

Topics:

  • Dataset preparation for detection
  • Annotation formats
  • Training detection models
  • Inference pipelines
  • Model evaluation (mAP)
  • Deployment considerations
tutorial
advanced

Ultralytics YOLO 11 Documentation

8-10 hours

Evaluation Metrics for Object Detection​

Understanding Detection Performance:

article
intermediate

mAP (Mean Average Precision) Explained

1-2 hours
guide
beginner

YOLO Performance Metrics

1 hour

See Data Science track for more metrics →

Capstone​

Apply your computer vision knowledge to a real-world project: Pokemon Card Border Detection System. Build a complete pipeline from data collection to deployment.