Portal Resources

The FathomNet Portal helps users train their own machine learning models by providing intuitive interfaces for annotation, streamlined access to the necessary computational power, and accessible tools to analyze the output.

Chrome and Edge are the recommended browsers as of Beta3 with limited support for Safari. The website styling is not targeted for mobile users, and test coverage has not included mobile as a primary use case. Let us know what you think! As our beta roll outs we are working on supporting a wide variety of features in a limited set of browsers reliably, and expanding the supported browsers as we grow. Currently authentication will work in most browsers, but users may not be abl

Learn how to organize your content more effectively by setting up nested folders and moving media files into folders you have created. Folders At the top of every project is a root folder, this is denoted as "/". All projects start with this initial root folder and it cannot be edited. When Media is in a folder, you can now view this media group in the main gallery view, and additional views provided you have the correct permissions, such as the Observations Gallery view. The

When you are ready to start adding media from a local computer to the portal via the website's upload capabilities, here are some screenshots and tips that can help you on your way. Also, learn more about transcoding. There are two ways to begin a web upload. First, you can choose single or multiple files from your computer file picker by clicking "Upload Files" in the media gallery, or you can drag and drop single or multiple files into the gallery. Subsequent the file selec

Granular user roles are a good way to introduce people into your project, while controlling the way they can interact with data. While each project team will have different requirements for data review, quality assurance, and quality control, the FathomNet team has done user research to identify these basic roles for the current version of Portal. Available user roles and descriptions Project Admin - Can edit all aspects of the project.   Project appears on Dashboard. Can up

Users on the portal can control who has access to their data using project privacy settings in addition to user management. Project Creation As a new user to Portal, if you haven't been assigned a role in a project, you will need to start one. As the project creator, your user will also be the admin of that project. You will give the project a name, a default avatar, and choose the privacy level. Project Privacy There are three privacy levels in the FathomNet Portal: Public

Advanced filtering enables users to create a custom query to display gallery listings of media and observations in the FathomNet Portal....

A more advanced way to work with media is connecting the FathomNet Portal to cloud or remote storage outside of the Portal. While the...

Portal supports generating datasets in formats popular with the AI community such as YOLO formats for Detection and Classification. This...

This tutorial focuses on leveraging Python and a custom utility to streamline the process of handling large files for Ocean Exploration and Video Work.

Analysis can be run on these medias without uploading and hosting on the portal, and the results exported. If you have your own hosting,...

Portal is an ideal location for users wishing to curate, and stage their data for other systems, including the FathomNet database....

Assembling an export from the portal starts with marking the items you wish to export. Marked items are either a still image, or a...

Portal algorithms can follow objects through time and generating annotations for a given object. The Observation Timeline allows for...

It is important to constrain labels so spelling mistakes, or capitalization differences don't create messy data. To achieve this, Portal...

The observations gallery allows verification and classification to be done upon many observations at once which can speed the process of...

Formatted sensor data can be imported and synced to media on a frame-by-frame basis. Go to an individual media's Related data tab to...

The algorithms in portal are listed in more detail in the Algorithm Library. This page aims to give an overview of where to launch...

Curating annotations is key to building supervised computer vision models, but it’s not without challenges.