Advanced Applied Computer Vision
Categories:
Artificial Intelligence
Reviews:
Currently Deep Learning is the most exciting field of Machine Learning. Deep Learning algorithms are giving state of the art result in almost every domain like computer vision, natural language processing, speech analysis, robotics etc.. This deep learning course is designed to introduce deep learning to students from basic to advance. After Completing this course students will be able to design the Deep Neural Network architecture for various application.
COURSE DESCRIPTION
How does the computers recognize objects, people, actions, animals, places, etc from images? This is a trivial task for humans but remained one of the core problems in Computer Vision. Recent advances in representation learning using multiple layers of abstraction (deep learning) have demonstrated to be an important aspect for designing artificial systems for visual recognition. In this class we will study and implement deep learning models and learning algorithms for visual recognition. After this class you will be able to understand, design, implement, and assess the impact of deep learning techniques for a diverse range of visual recognition tasks.
LEARNING OUTCOMES
-
- Understanding foundational concepts for representation learning using neural networks
- Become familiar with state-of-the-art models for tasks such as image classification, object detection, image segmentation etc.
- Obtain practical experience in the implementation of visual recognition models using deep learning.
The intended audience for this course:
- Data Engineers
- Data Scientists
- Machine Learning Engineers
- Integration Engineers
- Architects
Day-1
- Introduction
- Image Processing and Image Manipulation: Convolutions (Low Pass and High Pass Filters)
- Image Features: Gradients, HoG, SIFT, GIST, Bag-of-Features
- Introduction to Neural Networks (Deep Learning)
- Backpropagation and Optimization Methods
Labs
- Image Processing
- Features Detection from Images
- Multilayer Perceptron
- Machine Learning for Classification
- Gradient Decent Implementation
Day-2
- Convolutional Neural Networks (CNNs)
- Hyperparameter Tuning
- Regularization (Dropout, Augmentation etc)
- CNN Architectures I: LeNet, AlexNet, VGG
- CNN Architectures II: GoogLeNet, ResNet, DenseNet
- Object Detection I: RCNN, Fast-RCNN, Faster-RCNN
- Object Detection II: YOLO, SSD
Labs
- Convolution Neural Network
- ResNet, DenseNet
- Object Detection and Tracking(YOLO)
- Creating Dataset of object Detection
- Annotating Objects in Image
Day-3
- Image Segmentation: Fully-Convolutional Networks, Mask-RCNN.
- Image Synthesis: Style Transfer
- Variational Auto-Encoders (VAEs)
Labs
- Preparing Data for Mask RCNN
- Mask RCNN
- Style Transfer
- Image Captioning
- Model Deployment
Participants should have the following knowledge:
-
- Intermediate experience with Python
- Knowledge of Machine Learning Concepts (e.g. Overfitting) & Deep Learning Concepts (e.g. Backpropagation)
- Familiarity with Calculus (e.g. Multivariable Derivatives)
- Linear Algebra (e.g. Matrix Multiplication)