Add the power of computer vision to your university curriculum

Our team works hand-in-hand with your faculty to provide a modern curriculum with cutting edge tools for computer vision

Contact Us

Cutting Edge tools for a
modern curriculum

Overview

Omnis Labs is a technology startup based out of South Korea, the most innovative country in the world according to Bloomberg, and we have developed Deep Block as a suite for computer vision analytics. We are currently seeking partners in higher education who we can support by delivering courses and workshops to teach their students how they can use no-code tools to perform data-driven analysis from aerial imagery.

Our platform is suitable to develop courses for students and faculty who have identified particular problems in urban areas, such as traffic law violations and compliance with construction laws, to apply highly-scalable, automated analysis of datasets that include large aerial images. We intend to work alongside interested faculty to identify use cases and models that can be deployed for practical uses. Through courses using Deep Block, students will gain the skills for jobs in the knowledge economy, improving their analytical skills and equipping them with strong foundations for careers where they employ AI-driven tools to offer high-level analysis of real problems of human concerns.

city-street_t20_b473EB

Example Course

The following course is currently being piloted with leading universities in South Korea and in the United States. The general version of this course will also be available for individual learners online. 

Course Title:

Aerial Image Analytics for Urban and Environmental Insights

In this course, students learn the fundamental concepts and workflows involved in deep learning for computer vision. The course will consist of conceptual lectures introducing the basic principles of a data-driven analysis using aerial imagery, and each lecture will have a practical component where students will apply the principles using a point-and-click, no-code environment in a web browser.

 

 

 

No-code

This course has been developed for students in fields other than technology and computer science, so that they can use no-code tools.

Theory and Hands-on experience

The basic course consists of ten weekly lessons that include both theory and hands on practice of automated analysis using images. 

AI-driven analysis

to conduct AI-driven analyses using images from their own domain of research for commercial or for public sector concerns.

 

Practical Application

 Sudents will learn to find and import datasets, annotating their data to train the deep learning models, evaluate their AI model performance, and deploy the models to provide real-world analytics.

List of Lectures

Each lecture will consist of 45 minutes of a conceptual introduction to a particular step in an image analysis workflow, followed by a 45 minute demonstration of how to complete that step using Deep Block, a no-code, browser-based tool. We may also introduce other open-source or commercial tools at the request of the participating faculty.

Lecture 1:

Introduction to Deep Learning

Lecture 2:

Applications of Deep Learning and AI

Lecture 3:

Introduction to Object Detection

Lecture 4:

Data Preparation, Data Preprocessing

Lecture 5:

AI Model Training

Lecture 6:

Evaluating the AI Models

Lecture 7:

Inference of AI Models

Lecture 8:

Deployment of AI Models

Lecture 9:

Limitation of AI, Optimization Tips

Lecture 10:

Final Project
(creating custom AI object detector using Deep Block)

How It Works

Pre-requisites

Although the content of this course has been developed without a need for prior experience or course pre-requisites in computer science or statistics, the final project will provide opportunity for students to develop and apply their own image analysis workflow to the problem of their choice. As such, students will greatly benefit from previous coursework and experience in evidence-based analysis to identify problems for which a solution can be formulated by identifying visual features in aerial photography of the particular region in question.

 

Technical Requirements

 The course will be delivered online through lectures and through a browser-based tool for imaging analytics. Students will need access to a PC or Mac running a current version of Chrome (versions 88 or 89 as of March 2021). A broadband internet connection with at least 10 mbps downstream is required to view the video lectures. A high-bandwidth connection of at least 10 mbps is recommended to upload imaging datasets for the students’ analyses of their own imagery. In the event that students choose to import images from open data sets obtained from the internet, they should have sufficient hard drive space for temporary storage on their own computer.

Assessments

The assessment methodology can be developed in conjunction with the participating faculty, and it will follow the general principles for outcome-based education, such as typical rubrics and methods to assess capstone projects in university courses.

Although the content of this course has been developed without a need for prior experience or course pre-requisites in computer science or statistics, the final project will provide opportunity for students to develop and apply their own image analysis workflow to the problem of their choice. As such, students will greatly benefit from previous coursework and experience in evidence-based analysis to identify problems for which a solution can be formulated by identifying visual features in aerial photography of the particular region in question.

Deployment Details

Upon deployment, we have to distinct licensing pathways for you to choose from: 

We can provide access on a per-student basis to the video lectures and Deep Block AI Suite according to their individual email address all hosted on the Deep Block web domain. The student’s course registration details could be shared by the university only by providing the enrolled student’s email address.

Alternatively, we can provide content and tools to be hosted on the university's own infrastructure. This would entail a single institution-wide license for the text and video content for the lessons that can be provided to be hosted on the university’s learning management system of choice.

For both alternatives, we will provide access to cloud storage and GPU compute resources on the Deep Block cloud. In general, 50 GPU compute hours are sufficient to train a model suitable for a course project at the university level.

Reach out to us!

Get in touch with our team to learn more. Enter your details below, and we will be in touch with you to schedule a phone call to discuss how we can help you introduce imaging analytics into your higher education curriculum.