We care about the impact our work has on the world around us. That’s why we want to ensure our work paves the way to a sustainable future. Using 🌍Data for Good is how we achieve this. 

Technology can help resolve sustainability challenges related to climate change and biodiversity conservation. That’s why we commit time and resources to ideas that are genuinely having a positive impact on our planet. 

R Shiny provides a great way to showcase the ability of data for good. One such example is our Mbaza Shiny App, complementing the Mbaza AI.

Mbaza AI

Mbaza AI is an open-source, free-use application for biodiversity conservationists. Appsilon created this tool for change, in collaboration with researchers at the University of Stirling and The National Parks Agency of Gabon (ANPN), as part of our Data for Good initiative.

Mbaza AI automatically, accurately, and rapidly classifies animal species in camera trap images or videos. And it does so using a state-of-the-art artificial intelligence (AI) model.

Our model can classify 3000 images per hour and is up to 96% accurate, using an average laptop without an internet connection. And best of all – it’s free to use!

Mbaza Shiny App

Complementing the Mbaza AI algorithm is an interactive data explorer interface – Mbaza Shiny App. The Mbaza Shiny App intakes the data from the AI model and allows for analysis and visualization in an interactive dashboard. Shiny Dashboards are an excellent tool for telling data-centric stories.

“The Mbaza Shiny App integrates with the Mbaza desktop app for camera trap image analyses and can be used to automatically calculate daily activity patterns of different animal species, create maps and calculate measures of relative abundance with no coding or statistical knowledge.” – Robin Whytock, PhD, former Postdoctoral Researcher at The University of Stirling

Improving the Mbaza Shiny App for visualizing data produced by our AI model, offered a great opportunity to leverage another aspect of our technical expertise – R Shiny development. 

By combining our data science skills across AI & R Shiny programming, we found a way to use technology to accelerate nature conservation efforts in Gabon. In short, we used technology to fight threats to biodiversity and we made it available to you.

Challenges

Projects like this require a good understanding of the end user’s needs. In this case, our primary users are researchers, ecologists, and park rangers. These users are working tirelessly in the field and typically do not have extensive programming knowledge. 

This is just one side of the equation though. Product development requires knowledge exchange, flowing both ways. We as experts and consultants in our tech niches have to explain in a comprehensive way what is feasible or not. And our clients must share their domain knowledge. 

Are you looking to apply AI to nature conservation? Discover the technology helping endangered animals.

When exchanging domain knowledge and working with specialists, it’s important to overcome the following challenges:

  1. Succinctly express and understand the ultimate goal of the project (biodiversity conservation) and all intrinsic aspects of the end users’ (wildlife conservationists) day-to-day responsibilities.
  2. Address working with unique data. This was the first time we were tasked with visualizing camera-trap data, and due to the remote nature of the project access to end users was limited.
  3. Explaining R/Shiny strengths and limitations to manage expectations.
  4. Balancing all aspects of the work (time, quality, and functionality) so that both parties were satisfied with the end product.

In doing so, the end product will better serve the users and the project timeline will be realized.

Approach

Close collaboration rooted in open communication and mutual trust set the stage for overcoming these challenges. We discussed all requirements and pain points from our partners, and the whole team put themselves in the shoes of the end users. 

This is how we approached the implementation:

  1. Constant feedback from the partner – closely listening to the experts.
  2. Incorporate data and algorithms prepared by biologists into our application – combining domain knowledge and new technologies.
  3. Sound planning and robust quality assurance process throughout the entirety of the project.

You can apply this approach to any Shiny development project to build higher-quality products that are more likely to see user adoption.

Results

Today, the Mbaza Shiny App has progressed with improved functionalities. We’ve accelerated effective biodiversity conservation in Gabon, by speeding up the data analysis process. The user base is growing and user feedback is overwhelmingly positive. 

And at the same time, we know it is just the beginning. We want to reach more people and more projects across the globe. We want to help them efficiently protect endangered animals with the power of AI and R Shiny!

We’ve learned a lot from our fruitful journey with the Mbaza Shiny App. In many ways, we improved ourselves as well as efforts in biodiversity conservation:

  1. The team expanded our technical skillset in utilizing technology for biodiversity conservation.
  2. We gained a sense of fulfillment in delivering a useful tool for nature and society.
  3. We see the project continuing and expanding into something even greater.

Both we, and the clients, came away satisfied with the end product. You can explore the Mbaza Shiny App demo

Summary

“These tools greatly enhance the ability of protected area managers to rapidly analyze camera trap data and make informed conservation decisions. The Shiny app is in the development phase but we hope to roll it out soon as a core part of the Mbaza image analysis pipeline.” Robin Whytock, PhD, former Postdoctoral Researcher at The University of Stirling

If you’re looking to accelerate your biodiversity monitoring and conservation efforts with an R Shiny dashboard, check out Appsilon’s free-use Shiny Dashboard Templates. Simplify and speed up your Shiny dashboard build with our ready-to-use templates. The bundle contains several beautiful and easy-to-use templates with a range of features and tech stacks. The best part is – it’s entirely free!

Discover what it means to be an RStudio Full Service Certified Partner and how we can serve you.

If you need a more customized, advanced option – reach out to us. We’re here to help. Appsilon is an RStudio Full Service Certified Partner. We develop advanced R Shiny applications for Fortune 500 companies across the globe. We’d be happy to help you choose the right options for your use case. Let’s talk and see how Shiny can help you grow.

Tired of manual data labeling and mislabeled training data? See how the Appsilon ML team built a streamlit widget for cleaning ML labels.

This article was co-authored by Appsilon Project Manger Konrad Żurawski and D4G Lead Andrzej Białaś.