Duration
Advanced Deep Learning for Practical Computer Vision
3 days
Overview
The area of Natural Language Processing (NLP) is a subarea of Artificial Intelligence that aims to make computers capable of understanding human language, both written and spoken. Some examples of practical applications are: translators between languages, translation from text to speech or speech to text, chatbots, automatic question and answer systems (Q&A), automatic generation of descriptions for images, generation of subtitles in videos, classification of sentiments in sentences, among many others! Learning this area can be the key to bringing real solutions to present and future needs!
Equip yourself and your team with the knowledge needed to thrive in this AI-driven era.
Outline
Human Emotions Detection
- Data Preparation
- Modeling and Training
- Data augmentation
- Tensorflow Records
Modern Convolutional Neural Networks
- Alexnet
- Vggnet
- Resnet
- Coding Resnet
- Mobilenet
- Efficientnet
Transfer Learning
- Leveraging Pretrained Models
- Finetuning
Diving into the blackbox
- Visualizing intermediate layers
- Grad-cam Method
- Ensembling and class imbalance
- Ensembling
- Class Imbalance
Transformers in Vision
- Understanding VITs
- Building VITs from scratch
- Finetuning Huggingface Transformers
- Model Evaluation with Wandb
- Data efficient transformers
- Swin Transformers
Model Deployment
- Model Conversion from Tensorflow to Onnx
- Understanding quantization
- Practical quantization of Onnx model
- Quantization Aware training
- Conversion to Tensorflow lite model
What is an API
- Building the Emotions Detection API with Fastapi
- Deploy the Emotions Detection API to the Cloud
- Load tesing the Emotions Detection API with Locust
Object Detection with YOLO algorithm
- Understanding object detection
- YOLO Paper
- Dataset Preparation
- YOLO Resnet
- YOLO Loss
- Data augmentation
- Testing
- Data generators
Image segmentation
- Image Segmentation - Oxford IIIT Pet Dataset
- UNET model
- Training and Optimization
- Data Augmentation and Dropout
- Class weighting and reduced network
People Counting
- People Counting - Shangai Tech Dataset
- Dataset Preparation
- CSRNET
- Training and Optimization
- Data Augmentation
Image Generation
- Introduction to Image generation
- Understanding Variational autoencoders
- VAE training and digit generation
- Latent space visualizations
How GANs work
- The GAN Loss
- Improving GAN training
- Face generation with GANs
Programme scope, delivery arrangements and suitability are confirmed during enquiry.