You're pledging to donate if the project hits its minimum goal and gets approved. If not, your funds will be returned.
Peak Finder: Offline Mountain Identification for Nepal
An offline mobile app using location data and computer vision to identify and explore Nepal's mountains.
Since my school days, I have had several memorable arguments with my friends about identifying mountains. We would see a mountain from a distance and argue about which mountain it was. Those small moments became some of my core childhood memories.
Now that I have moved towards the field of technology and AI, I started wondering if I could build something that could actually identify those mountains through a phone, even without an internet connection.
That is the idea behind Peak Finder.
Peak Finder aims to be an offline mobile application focused on identifying the peaks of Nepal. The main approach will use location-based and elevation data to determine the mountains that could be visible from a user's location. As a separate feature, I also want to develop a computer vision model that can identify a mountain from a camera image without depending on elevation data.
The app will also work as a mountain information platform, containing information and images of mountains across Nepal. Users will be able to identify a mountain, capture its image, save it with the mountain's identification, and share it with others.
The main goal of Peak Finder is to make mountain identification simple and accessible, especially for people traveling through Nepal.
The project will focus on the following goals:
Build an offline mobile application that can be used without a continuous internet connection, making it useful even while traveling through remote areas.
Develop a location and elevation-based identification system that uses the user's location, elevation, and mountain data to determine which peaks they are likely looking at.
Develop a computer vision model that can identify mountains directly from a camera image. This will be a separate approach from the location and elevation-based system.
Create a mountain database for Nepal containing mountain names, elevations, locations, images, and other useful information.
Make mountain identification useful for tourists and mountain lovers, allowing them to quickly understand what they are seeing while traveling through Nepal.
Allow users to capture and save mountain memories, where a photograph can be saved together with the identified mountain and its information.
Allow users to share their identified mountain photographs, making the application useful not only for identification but also for exploring and sharing experiences.
To achieve these goals, I plan to collect and prepare the required location, geographic, and elevation data for Nepal's mountains. I will also collect available mountain images and prepare them as a dataset for training and testing the computer vision model.
Alongside the AI component, I will gather information and images about Nepal's mountains and organize them into a database that can be used by the application.
The different components will then be integrated into a mobile application with offline usage as one of the main considerations.
The funding will mainly be used for developing the mobile application, collecting and preparing the data required for mountain identification, and training and testing the computer vision model.
It will help with collecting mountain images, preparing geographic and elevation data, developing the mountain database, and providing the computing resources needed to train the AI model.
The funding will also support testing the application and making sure the different components can work together effectively in an offline environment.
Dinisha Uprety — UI/UX Designer and Product Manager
Our team has experience developing AI and computer vision applications for practical use cases.
We have previously developed a computer vision model that allows users to capture images of jaws and automatically detect relevant dental conditions. This application is being developed for use in dental camps.
In the dental camp system, students' images can be captured and then reviewed and annotated by doctors. The doctor can provide the necessary recommendations and next steps, after which a report can be generated and sent digitally to the student's phone.
The idea is to provide students with a safe digital record and better awareness instead of relying only on traditional paper-based reports, which can easily be misplaced after a dental camp.
This previous project has given us experience with computer vision models, image data, AI-based detection, application development, and building technology around a real-world problem. We want to bring this experience into Peak Finder and apply it to mountain identification in Nepal.
The biggest risk for this project is the availability and quality of data.
Collecting enough reliable elevation, location, and mountain image data could take more time than expected. The computer vision model will also depend heavily on the quality and variety of the images available for training.
Mountain identification from photographs can be difficult because the appearance of a mountain changes depending on weather, lighting, distance, viewing angle, clouds, and season. Some mountains can also look similar from certain locations.
If the computer vision model does not achieve the expected accuracy initially, the location and elevation-based identification system can still provide a useful foundation for the application. The data collected and the work done on the application can also be used to continue improving the model in the future.
None till now