Design, train, and deploy computer vision models
This intensive training course provides an understanding of and teaches how to implement convolutional neural networks (CNNs), which are widely used in image processing and computer vision.
Is it for you ?
AI Developers / Data Scientists, Computer Vision Engineers
Prerequisites
• Basic knowledge of Python
• Understanding of machine learning concepts (regression, classification)
• Understanding of matrices and linear algebra
What You'll Walk Away With
- ✓ Build and train CNN models for image classification using Keras
- ✓ Prepare and enhance datasets with normalization, augmentation, and annotation
- ✓ Optimize model performance using advanced architectures and hyperparameter tuning
- ✓ Evaluate and compare models with appropriate metrics and validation techniques
- ✓ Deploy computer vision solutions using transfer learning on real-world use cases
Training content
1 Fundamentals of CNNs
Chapter 1: Introduction to convolutional neural networks
- History and applications
- Comparison with traditional networks (MLP)
- Basic architecture: convolution, pooling, activation
Chapter 2: Data preparation
- Image formats and normalization
- Data augmentation
- Annotation and standard datasets (MNIST, CIFAR, ImageNet)
Workshop 1: Building a first CNN with Keras
- MNIST image classification
- Visualization of layers and filters
2 CNN model design and optimization
Chapter 3: Advanced architectures
- VGG, ResNet, Inception
- Deep networks and overfitting issues
- Dropout, batch normalization
Chapter 4: Training and evaluation
- Cost function and metrics
- Cross-validation
- Early stopping and hyperparameter tuning
Workshop 2: Implementing a CNN model on CIFAR-10
- Comparing multiple architectures
- Performance analysis
3 Practical applications and transfer learning
Chapter 5: Detection and recognition
- Object detection (YOLO, SSD)
- Facial recognition
- Image segmentation
Chapter 6: Transfer learning and fine-tuning
- Using pre-trained models
- Adapting to specific datasets
- Advantages and limitations
Workshop 3: Final project – Object detection in real images
- Using pre-trained MobileNet or ResNet
- Deploying a simple model
📌 Practical information
Our training sessions are offered in Montreal or Quebec City, in person or in a virtual classroom. Dates and locations are specified when you select your session below. If you have any questions, check out our FAQ.
