A cross-platform mobile app built with Flutter/Dart for users of Soundbendor Lab's Smart Compost Bin.
The binsight.ai app enables Wi-Fi connectivity for the Smart Compost Bin so that participants in a field study will be able to use their bin to track household food waste, annotate detection images, and follow their compost data trends over time.
- Repos
- Purpose
- Value Proposition
- Getting Started
- Usage
- Features
- Documentation
- Contact Us
- Contributions
- License
- Food Waste: https://github.com/Soundbendor/food-waste
- Food Detection embedded: https://github.com/Soundbendor/food-detection-embedded
In the face of an escalating climate crisis, finding more effective food waste disposal methods remains a pressing challenge in the field of sustainability research. To contribute valuable data to efforts in this field, the binsight.ai team pioneers one approach that is aimed at expanding our understanding of food and organic waste at the household and consumer level.
Our project supports a countertop Smart Compost Bin equipped with sensors that capture images of food waste and many associated data points. These include bin internal temperature and humidity, IAQ, CO2, the weight of composted food items, thermal imaging depth maps, and VOCs.
To serve as the user interface to the smart compost bin, we developed a cross-platform mobile application. This user-friendly app tracks usage and informs users about their composting habits. When a new user opens the app for the first time, it guides them through a one-time setup process, using a Bluetooth pairing routine to connect their smart compost bin to the user’s Wi-Fi network. Once connected, the app serves as a portal that provides a data-driven view that tracks a user’s composting habits.
Each time the user opens the app, it automatically downloads any recent images of that user’s compost from our database server. For each new image, the app tasks the user to draw an outline around each compost item in the image using their touchscreen. We save this boundary annotation alongside the transcribed categorical label to our database. These annotated boundaries allow us to train a robust semi-supervised segmentation model to automatically detect food waste. We will deploy this app for our initial data collection field study to build a dataset of labeled and manually segmented examples. This dataset supports the training of our automatic food waste identification model, and future versions of this application will simply perform food waste detection automatically, without requiring user input.
Additionally, we provide an interactive analytics dashboard to inform users about their composting habits. This dashboard highlights common behavioral trends, including graphs illustrating their most frequently discarded food items. Our mobile app provides personalized analytics to the user with the overarching goal of encouraging informed decision-making that leads to reductions in their food waste footprint.
This app enables crowd-sourcing the laborious task of creating segmentation maps to label the individual items present in the commingled food waste. We acknowledge that users will not necessarily exhaustively annotate every single item they discard, but over time we will curate a large collection of labeled items. Furthermore, because we deploy our smart bin and app in a field study and we do not restrict the compostable food items users are allowed to discard, we will curate a large dataset better representative of food items that users send to compost.
The binsight.ai team is pioneering a new way in the field of Smart Composting and image annotation, and as such, there is no direct competition. All current solutions that are out there are working towards the same goal: Fighting the effects of food waste and climate change.
- Clone the repository to your local machine:
git clone https://github.com/Soundbendor/smart-bin-app.git-
Install the Flutter SDK on your machine. Follow the instructions here.
-
Build and run the app on your local machine:
flutter runSee the DEVELOPMENT.md file for more information on how to develop and debug the binsight.ai app.
To use the binsight.ai app:
- Ensure you have a functional Wi-Fi network, as well as Bluetooth capability on your desired mobile device.
- Install the app on the mobile device of your choice.
- Open the app and complete the bin set-up process.
- Use the Smart Compost Bin as you would any other composting solution.
- After composting an item, check the detections page on your app to access and annotate your detection.
- Watch your composting stats change on the dashboard over time.
- Repeat from step 4, and have fun!
The binsight.ai app offers the following features and functionalities:
- Wi-Fi credential sharing via Bluetooth bridge to successfully connect the bin to its MongoDB Cloud database.
- Image annotation tools for users to draw object outlines and label food waste to build a novel dataset.
- A dashboard with data visualizations and analytics to support food waste goals, and increase user engagement.
The documentation for the binsight.ai app can be found in the docs directory.
The documentation can also be found through GitHub Pages here.
Please feel free to reach out to our support email at:
Current contributors to this project are as follows:
Advising Professor:
- Dr. Patrick Donnelly, Oregon State University-Cascades
Development Team:
- Blake Pugh, OSU student
- Daniel Lau, OSU student
- Kimberly Markley, OSU student
- Micah Stahlberg, OSU student
See the LICENSE file for details.