We use cookies to optimize your user experience. We also share information about your use of our site with our social media, advertising and analytics partners. By continuing to use our site you agree to use cookies in accordance with our Privacy Policy.

Bringing Innovation to help the environment with Fishial.AI

At Codahead “innovation” is not just a buzzword. We are always striving to use the latest technological advances to help our clients reach their goals. If by doing that we can also help the environment, as is the case with Fishial.AI, we are even happier to be a part of the project.

Conceived by The Wye Foundation, Fishial.AI seeks to be at the forefront of technology to assist in global conservation efforts. As Codahead, we are building a portal that will allow the public to participate in the following: uploading images, creating polygons around fish in images, tagging species and identifying attributes, within their own images or those of other users. The idea is to have people all around the world help create the largest training dataset for an AI model for fish species recognition. The resulting trained model will be open sourced offering endless opportunities to help the scientific community, fishing community and fishermen worldwide. With the potential to be incorporated into any platform, scientists, fishery managers and fishermen can create platforms that best suit them to operate fishial recognition powered by Fishial.AI.

We want the model to take advantage of modern neural networks, so instead of tagging fish with standard bounding boxes (which include the background and hence introduce noise to the training process), we opted for polygons which enable pixel-based semantic segmentation of the image.

Tagging with polygons tends to be very time consuming, as users need to meticulously drag every point of the polygon to make sure it fits the object on the image as precisely as possible. As we wanted the user experience with the portal to be fast and streamlined, we were able to incorporate Computer Vision related algorithms even at this stage of the project. The resulting flow can be seen on the image below — it requires just a few user clicks to precisely mark the fish on the image.

The word is spreading and Fishial.AI is gaining more attention. The project was recently featured by ConservationXLabs and is now one of the 20 finalists that are competing for the main prize. Be sure to follow https://fishial.ai to be notified when the portal is live so that you too can participate in building the largest fish recognition AI database in the world!

Also, if you are interested in what we do at Codahead and how we can help your business grow by implementing innovative software solutions, feel free to contact us at sales@codahead.com.


Fishial.AI: https://fishial.ai
Fishial.AI at ConXTech: https://conservationx.com/project/id/350
Codahead: https://codahead.com



Bartosz Fijalkowski

Ruby on Rails, AWS Dev ops