Meet people from around the globe?
Get a great reference letter to support your job applications
We would like help to get more help on the
Observations Revisited Project
NHI Revisited Project
The Blue Box image processing project
The Blue Box Mini station
The Field Trip Week Planning
We would like help to get your help on the
Hypothesis Project
Experiments Project
We would like someone to create a page with links to information about weather, solar storms etc.
We would like someone to put all the observations pages from the old hessdalen site into an updated Google Document - and do some gemini magic to extract some information (colors and shapes observed, etc.)
We would like someone to evaluate and suggest improvements to our Discord channel.
We would like someone to suggest design elements to use on the web and for printed material.
We would like someone to create TikToc / YouTube short videos ;-) We have the content you need!
We need someone to help us apply for funding and get donors!
We need help to get more members (to pay our internet bills)
We need more people with technical knowledge to build the new Blue Box Mini
React, Next
Copilot
Codespaces
Typescript
HTML, CSS
image processing
machine learning (ML)
Help us to build an
Anomaly Detection and Recording System
Drones with AI
and multiple sensors.
Help us to create an
automated drone monitoring system
Proposal for a Bachelor's Thesis
Automated Image Processing
Problem Statement
Project Hessdalen produces 90 GB of images every day. The majority of these images are completely uninteresting - so first and foremost, a function is needed to split the image material into two: uninteresting - and potentially interesting.
Behind each image lies a 20-minute film. If we can find the images that are potentially interesting, we can reduce data transfer to a fraction and simplify further work. The goal is to be able to transfer images and films of interest to a cloud solution for further analysis. These images can then be categorized: by using AI models that can identify what is in the image and the associated film. The images may contain birds, insects, airplanes, meteorites, satellites, and other things. For Project Hessdalen, the hope is that an UFO will appear in the "other" category! It is desirable to achieve a system that can identify what is in the image - primarily based on the stacked image created from the last 20-minute sequence. If possible, it is desirable to be able to use a system to analyze a real-time recording, so that it can be alerted instantly if an event of interest is detected, i.e., something that falls into the "other" category.
Context and Background
This task requires understanding of AI modeling, but also communication between machines. Being able to process either locally or as a service on a virtual machine is important - it is important to figure out the right configuration: For example, if it is desirable to reduce traffic over a 4G network, local processing is important to provide rapid notification.
There are tools available that can be used - but the task is interesting because data that needs to be processed is created daily.
We currently have 2 cameras - and each camera produces 45 GB of data.
The hope is to be able to scale up to many more cameras - but this must be handled in a better way.
It is important to solve this task because everything is currently stored locally. Backups are taken - but the amount of data is enormous. 45 GB x 2 cameras x 365 days = a lot of data every year!
The system has soon been operational for one year.
Proposed Solution or Approach
All stacked images are stored on a Google Drive.
A system can be created that reads directly from this, or one can download a few days' worth of image material and test a solution locally.
The system must then be able to be installed on the computer where the cameras are installed, so that local processing can be done before data transfer.
The machine uses Linux as its operating system.
It is desirable that the system should be able to work in a cloud service, locally (Linux), and preferably on a Raspberry Pi - to reduce the costs of a local machine.
The system will primarily be installed in Blåboksen (The Blue Box) up in Hessdalen.
Scope and Limitations
It should be possible to concretize tasks that can be solved within six months.
Then, a series of tasks can be linked together to achieve a complete solution.
There can be different focus areas that can be worked on in parallel:
Training an AI model
Automatic transfer of data and cleanup after transfer
Notification mechanisms
Sending out mobile notifications
Sending out emails with details about alerts
Real-time analysis - can an AI model be used to provide instant alerts?
Potential Challenges and Learning Outcomes
All the necessary data is available - the challenge will be to test the system:
The current solution should not be stopped when installing and testing whether the solution works.
The advantage of the task is access to data and a clearly defined need here, theory can be tested in practice. The knowledge and experience gained can easily be transferred to other areas: There is a great need for analysis of images or cameras - preferably in real-time - whether it concerns monitoring traffic, animals, fish, weather conditions, or studying celestial phenomena.
Resources and Availability
The project can provide simple equipment - such as a Raspberry Pi, but it is expected that students have a PC that they can use.
The project will create GitHub repositories for sharing code.
Students will receive their own account for use and access to Google Drive - where all data is stored.
There are several team members who can be available to students if needed, and I will personally be available when guidance is needed.
Sincerely,
Fred Pallesen
CEO of Project Hessdalen
Proposal for a Bachelor's Thesis
Knowledge Graph with AI
Problem Statement
One goal for Project Hessdalen is to document and analyze how the Hessdalen phenomenon has been perceived, interpreted, and described over time, with a particular focus on local accounts, archival material, and previous research.
The result should be an accessible and living knowledge base that can be used further in research, dissemination, and exhibitions.
Context and Background
The project has many previous research reports, scientific articles, and other material that needs to be analyzed and contextualized - as part of the project's continuous knowledge base.
All documentation – including archival findings, interviews, and analyses – needs to be digitized, tagged, and systematized in a structured database.
Proposed Solution or Approach
A knowledge graph will be built, which will make it possible to link events, people, interpretations, and observations across time and sources.
In the long term, the database will be connected to an AI-based dialogue model, which will allow researchers, youth, and the general public to "talk to" the knowledge base – ask questions, retrieve information, and explore connections in an intuitive way.
Scope and Limitations
It should be possible to concretize tasks that can be solved within six months.
Then, a series of tasks can be linked together to achieve a complete solution.
There can be different focus areas that can be worked on in parallel:
Modeling a knowledge graph
Creating import functions that automatically add documents to the knowledge graph
Connecting to an AI that can use the knowledge graph and the documents when having a dialogue with the model
Potential Challenges and Learning Outcomes
All the necessary data is available. All the tools and methods needed exist
the challenge will be to set up and document the system - and make it easily accessible. The advantage of the task is access to data and a clearly defined need
here, theory can be tested in practice. The knowledge and experience gained can easily be transferred to other areas: There is a great need to be able to create domain experts that can be used by people with different levels of knowledge about a topic. Being able to create a system that grows as new articles, observations, and data become available is also important to ensure an up-to-date system.
Resources and Availability
The project can provide access to the necessary internet services.
It is expected that students have a PC that they can use.
The project will create GitHub repositories for sharing code.
Students will receive their own account for use and access to Google Drive
where all data is stored. There are several team members who can be available to students if needed, and I will personally be available when guidance is needed.
Sincerely,
Fred Pallesen
CEO of Project Hessdalen
Want to have a fun project in your (soon to be gone) spare time? We are looking for motivated members in the following fields:
Embedded Developers
Program Arduinos, ESPs or your favorite microcontroller to acquire data in the valley.
Sensor Experts
You know how to I2C or SPI? You know the registers of the HMC5883L by heart and know that RM3100 is in a totally different league? Call us! All other sensor experts are also appreciated :-)
SDR, ADSB, VLF, LoRa and Co
Operators for SDR, ADSB and the long-dormant VLF antennas can have an absolute marvelous playground in the Blue Box. It’s warm in winter and mosquito-free in summer!
Big Data or AI
You like Terabytes of data? Look no further - we have them. Months and months of video footage that can be used for existing and new detection systems.
Engineer & Electricians
We are building waterproof sensor boxes and bolting cameras to the pole. Hardware help is always appreciated.
You like tech AND animals - sheep cam!
Animal lovers: pursue our latest idea to get this definitive UAP picture we all hope for: put a camera on a sheep and see the world like your fluffy friend 🙂
Project Hessdalen is looking for volunteers to contribute to our website redevelopment. We're using Sanity and modern web development practices.
If you're interested in contributing to an open and educational project, please contact us at admin@hessdalen.org.
The old.hessdalen.org is still available - created by handwritten HTML.
The new hessdalen.org is built using Google Site. It is an easy to use website builder, but lacks the robust design and dynamic content features of platforms like Sanity.
Sanity excels for:
Flexible content modeling: Tailor content structure precisely.
Structured content: Enables powerful, reusable data.
Developer-friendly: Integrates seamlessly with modern tools.
Real-time collaboration: Streamlines content workflows.
Sanity empowers content creators by providing them with a flexible and intuitive platform to manage their content effectively.