💡 Inspiration

Walking alongside the streets in our neighborhood, we can't help but notice the beautiful variety of plants thriving in our urban ecosystem. Beneath this seeming affluence, however, the danger of non-native species--or invasive species--cannot be ignored. Indeed, the introduction of alien species into a new ecosystem may lead to economic, environmental, and human health damages. Research suggests that these invasive species compete with native species for nutrients, space, and sunlight, often leading to soil erosion, decreased biodiversity, deteriorated water quality, and poorer agricultural lands. According to the US Department of Agriculture (USDA), invasive species have contributed to the decline of 42% of U.S. endangered and threatened species, and invasives are in fact the main cause for 18% of those declines.

As high school students, we believe this is a vital yet largely marginalized issue that demands closer inspection by society. We, therefore, decided to use this hackathon as an opportunity to combat the crisis of growing invasive species across our communities, and we took on an innovative approach to this problem: EcoDefender.

📷 What it does

EcoDefender combats the issue of invasive species by encouraging the general public, rather than just government officials, to participate in the process of identifying and reporting potential invasive species. EcoDefender is extremely easy to use. First, users open either our web app or the app installed on their phone and upload a photo of potential invasive species. Our machine learning algorithm will process the image and identify the type of species shown in the picture, and, depending on the location of the user, determine whether this particular species is considered invasive. Often, the general public lacks the needed background knowledge to identify a non-native invasive species, and it's unrealistic to assume that trained officials alone are enough to reduce the adverse impacts of the invasives. Our platform allows the public to detect possible invasive species even without proper training, drastically increasing the effectiveness of our initiative.

Next, users can report their findings to city officials by submitting relevant data on our app. Subsequently, their reported data will be combined with geofencing software that will flag all places with reported invasive plants. This allows officials to closely examine the possible patterns in the distribution of invasive species and determine their potential origins. EcoDefender also includes explanations of what an invasive species is, further encouraging the general public to take part in this environmental initiative. We believe an easy-to-use platform like EcoDefender would greatly galvanize the general public into helping find and remove invasive species.

💻 How we built it

After hours of brainstorming, research, and planning, we decided to split our work by assigning Chris to Figma app prototyping, Jesse to machine learning model development, and Jarvis and Alan to front-end development.

Overall, we trained the machine learning model for classifying species with Teachable Machine, obtaining a TensorFlow.js model that's incorporated on our website. The model is trained with around 250 different categories of invasive plants and animals, obtained from the Global Invasive Species Database. We also built the project using HTML, CSS, and JavaScript for the front-end and developed a prototype phone app using Figma. The website itself is hosted through Vercel.

For the complete code we used in the project, please refer to our GitHub repository linked below.

🚨 Challenges we ran into

As the backbone of our project is a machine learning model, obtaining a reliable and reasonably large dataset on different invasive plants was difficult. While there are no publicly available datasets, we managed to systematically download relevant images on Bing with a program.

In addition, we had to navigate through platforms that we are unfamiliar with, including Figma. We also had no prior experience with JavaScript, making the incorporation of TensorFlow model particularly challenging. Although these technologies were foreign to us at first, learning them throughout this hackathon was a truly rewarding experience, and we are glad that a rough prototype of our project is completed.

🚩 Accomplishments that we're proud of

We are very proud that we were able to apply machine learning to contribute to the environment around us. This is the first time we implemented computer vision technology to help the general public identify potentially hazardous species, and it was gratifying to see how a technology that appears aloof and impersonal can be utilized to create such a genuine positive change. We also take pride in the innovative approach we took to address the issue of invasive species.

As this is the first Hackathon for many of us, we are also proud of the relative completeness of our project. We successfully incorporated around 250 different invasive species, an incredibly large number given the limitation of a hackathon. Learning many foreign platforms and technologies, while experimenting with the coding languages we knew before, was overall a memorable and satisfying experience.

🏫 What we learned

In terms of software and design, we gained experience with web development and Figma, two aspects of computation that were alien to us before the Hackathon. This is an important learning experience that improves our collaboration and time management skills. We also learned more about the danger of invasive species to society through research. We encourage everyone to explore more about invasive species at the National Wildlife Federation.

🤠 What's next for EcoDefender

Because of the time constraint of the hackathon, we were unable to produce an actual app that can be installed on users' phones. In addition, the geofencing feature is only modeled but not realized in our project. Moving forward, we wish to develop an actual app for EcoDefender and include the map locations feature on our platforms. We also intend to refine our machine learning model with even larger datasets.

Furthermore, to bring about true changes, we also need to further advertise our product among the public and collaborate with local officials. Although EcoDefender is far from perfect now, we are confident that it will ultimately evolve into a true ecosystem defender, protecting our environment from alien species invasion.

Built With

Share this project:

Updates