fbpx
Learn to build large language model applications: vector databases, langchain, fine tuning and prompt engineering. Learn more

Create bird recognition app using Microsoft Custom Vision AI and Power BI

Data Science Dojo
Saumya Soni

December 15

Learn how to create a bird recognition app using Custom Vision AI and Power BI for application to track the effect of climate change on bird populations.

Imagine a world without birds: the ecosystem would fall apart, bug populations would skyrocket, erosion would be catastrophic, crops would be torn down by insects, and so many other damages. Did you know that 1,200 species are facing extinction over the next century, and many more are suffering from severe habitat loss? (source).

Birds are fascinating and beautiful creatures who keep the ecosystem organized and balanced. They have emergent properties that help them react spontaneously in many situations, which are unique to other organisms.

Here are some fun facts: Parasitic jaegers ( a type of bird species) obtain food by stealing it directly from the beaks of other birds. The Bassian Thrush finds its food using the most unique way possible: they have adapted their foraging methods to depend on creating a large amount of gas to surprise earworms and trigger them to start moving (so the birds can find and eat it).

Due to the intriguing behaviors of birds, I got inspired and lifted to create an app that could identify any bird that you are captivated by in real time. I also built this app to raise awareness of the heart-breaking reality that most birds face around the world.

 

Global trends of bird species survival chart

I first researched bird populations and their global trends from the data that contains the information of the past 24 years. I then analyzed this data set and created interactive visuals using Power BI.

This chart displays the Red List Index (RLI) of species survival from 1988 to 2012. RLI values range from 1 (no species at risk of extinction in the near term) down to 0 (all species are extinct).

As you click on the Power BI Line Chart you will notice that since 1988, bird species have faced a steadily increasing risk of extinction in every major region of the world (change being more rapid in certain regions). 1 in 8 currently known bird species in the world are at the threshold of extinction.

The main reasons are degradation/loss of habitat (due to deforestation, sea-level rise, more frequent wildfires, droughts, flooding, loss of snow and ice, and more), bird trafficking, pollution, and global warming. As figured, most of these are a result of us humans.

Due to industrialization, more than 542,390,438 birds have lost their lives. Climate change is causing the natural food chain to fall apart. Birds starve with lesser food (therefore must fly longer distances), choke on human-made pollutants, and end up becoming weaker. Change is necessary, and with change comes compassion. This web app can help to build an understanding and empathy toward birds.

Let’s look at the Power BI reports and the web app.

 

Power BI report: Bird attributes / Bird Recognition

As you can see in this report, along with recognizing a specific bird in real-time, interactive visualizations from Power BI display the unique attributes and information about each bird and its status in the wild. The fun facts on the visualization about each bird will linger in your mind for days.

AI web app – To create a bird recognition app

In this web app, I used cognitive services to upload the images (of the 85 bird species), tagged them, trained the model, and evaluated the results. With Microsoft Custom Vision AI, I could train the model to recognize 85 bird species. You can upload an image from your file explorer, and it will then predict the species name of the bird and the accuracy tied to that tag.

The Custom Vision Service uses machine learning to classify the images I uploaded. The only thing I was required to do was specify the correct tag for each image. You can also tag thousands of images at a time.

The AI algorithm is immensely powerful as it gives us great accuracy and once the model is trained, we can use the same model to classify new images according to the needs of our app.

  1. Choose a bird image from your PC
  2. Upload a bird image URL
  3. Take a picture of a bird in real-time (only works on the phone app as described later in the blog)

Once you upload an image, it will call the Custom Vision Prediction API (which was already trained by Custom Vision, powered by Microsoft) to get the species of the bird.

Bird recognition using AI
Measure the effect of climate change on birds

Phone application

I also created a phone application, called ‘AI for Birds’, that you can use with camera integration for taking pictures of birds in real time. After using the built-in camera to take a picture, the name of the bird species will be identified and shown. As of now, I added 85 bird species to the AI model, however, that number will increase.

The journey of building my own custom model, training it, and deploying it has been noteworthy. Here is the link to my other blog for how to build your own AI custom model. You can also follow along with these steps and use them as a tutorial: Instructions for how to create Power BI reports and publish them to the web will also be provided in the other blog.

 

Conclusion

The grim statistics are not just sad news for bird populations. They are sad news for the planet because the health of bird species is a key- measure for the state of ecosystems and biodiversity on planet Earth in general.

I believe in Exploring- Learning- Teaching- Sharing. There are several thousands of other bird species that are critical to biodiversity on planet Earth.

Consider looking at my app and supporting organizations that work to fight the constant threats of habitat destruction and global warming today.

Our Earth is full of unique birds which took millions of years to evolve into the striking bird species we see today. We do not want to destroy organisms that took millions of years to evolve in just a couple of decades.

Sources:

Newsletters | Data Science Dojo
Up for a Weekly Dose of Data Science?

Subscribe to our weekly newsletter & stay up-to-date with current data science news, blogs, and resources.

Data Science Dojo | data science for everyone

Discover more from Data Science Dojo

Subscribe to get the latest updates on AI, Data Science, LLMs, and Machine Learning.