Updated: May 10, 2020
There are many citizen science based apps and efforts that have been developed to address litter collection. This article is an attempt at consolidating a listing of these projects. The only criterion for inclusion in the article is that a project's structure makes it possible for anyone to contribute data to the cause from their current location. The projects mentioned thus far are chronologically ordered. A brief description of each project’s establishment, functionality, data, achievements, and contact info is provided. The list is not comprehensive, so if an effort has been left out, which meets the standard stated above, then please reach out to us here so we can consider it for inclusion.
The hope is that a consolidated list will help ensure a shared awareness of individual project efforts, incentivizing cross collaboration and the sharing of best practices. The end goal of the projects mentioned is a shared goal, elimination of litter from the natural environment, which hopefully motivates people to reach out and assist one another where able. Each one of these projects brings a somewhat unique spin on litter collection to the table with the potential to save time and effort by sharing what has already been done. :)
The Litter Platforms:
Pirika: is a startup that was created in 2011, at Kyoto University in Japan, with the goal of solving environmental problems using science and technology. It boasts itself as the most widely used smartphone litter app, with over 800,000 users across 100 countries and is available on both Android and iPhone. Pirika also offers free litter data visualization services for companies, organizations, and local communities to help invigorate cleanup efforts; over 300 companies have been assisted to date. Additionally, #Pirika has developed Takeanome, which is an artificial intelligence (AI) program that maps and measures litter using visual recognition technology. Takanome has mostly been used in Japan and the research helps assess the effectiveness of anti-litter governmental efforts. The data is captured via smartphone video and analyzed by AI.
Litter heatmaps are then generated depicting the density of litter throughout the areas researched and anti-liter design proposals generated upon requests. Pirika is supported by numerous companies, governments, organizations and to date has collected over 1.4 million pieces of litter; 95% of this litter having been collected in Japan. Pirika has both non-profit and for-profit components and is open to collaboration with universities and corporate research teams, but only makes portions of its data freely available to the public. The Pirika contact page can be used for questions, comments, and learning how to contribute.
Litterati: is a startup established by Jeff Kirschner in 2012, with the intent of crowdsourcing litter detection and collection via a smartphone app, similar to Pirika, which is available on both Android and iPhone; the app itself was funded via Kickstarter in 2017. The #Litterati app is unique in that it has AI built into it, LitterAI, to recommend tags when a piece of litter is imaged. The app also gamifies litter collection by allowing users to invite friends and create litter challenges. Over 5400 local challenges and 21 global challenges have been launched to date, which have led to cultural changes at both the company / brand and local levels. To date, Literati has over 162,000 users across 165 countries, which have accounted for nearly 5.5 million litter collected. The global litter count can be visualized on the Litterati map and the most tagged litter items visualized here. Litterati partners with cities, schools, citizen scientist, businesses, and NGOs via paid plans to help them measure the impact their teams are having on litter reclamation. Litterati is a for-profit company, and currently only allows individuals to download the data that they have contributed personally. However, Litterati’s FAQ page states that they plan to adopt a fully open data policy in 2020 after addressing the privacy concerns of certain contributors. For questions and comments reach out to Litterati via their support address.
OpenLitterMap (OLM): is a solo effort led by Seán Lynch, which was started in 2013; the app itself was self funded and launched in 2017. OLM is a Snapchat inspired app, like other apps mentioned above, used for taking geolocated pictures of litter and tagging it based on predefined litter categories presented to the user. Once labeled, the images are uploaded to the OLM cloud enabling patterns of litter deposition and type to emerge overtime. The images have yet to be employed in AI applications, but it is easy to see the potential use of this data archive, and its scalability as it enables every human to be a litter labeling sensor. Litter that has been imaged and uploaded over the prior week can be viewed on the OLM website, but the entire dataset is made available and free for use. OLM also gamifies litter collection by hosting an ongoing “World Cup” in which users are represented as members of their country who compete to be global OLM contributing leaders; the top 10 contributors are displayed on OLM’s “World Cup” page. Uniquely, OLM users can also earn #Littercoin for their contributions, which is a blockchain based reward. Currently more than 2,900 Littercoin have been issued, but there is no monetary value or utility associated with this reward yet, other than bragging rights. The OLM app is available on both Android and iPhone and the apps are constantly improving. However, the necessary manpower and funding are simply not there to provide updates as quickly as possible. Seán has applied for 60 grants and run a Kickstarter to support the OLM effort, all of which have gone unfunded. OLM truly is a one man show and it is amazing what Séan has been able to accomplish with minimal resourcing; over 152,000 pieces of litter have been geotagged via 72,000 user submitted photos. Currently #OpenLitterMap is run entirely based on user donations. For those interested in helping contribute to the development of the program please reach out to Séan via e-mail.
Clean Swell: was launched in 2016 by Ocean Conservancy and is a global movement to keep beaches, waterways and the ocean trash free. Clean Swell is an app, available on both Android and i-Phone, that allows you to catalogue each piece of trash you collect and share your results with friends via Facebook, Twitter, and e-mail. The data collected is instantly uploaded to the Ocean Conservancy’s global ocean trash database providing a global snapshot of ocean litter to decision makers to help inform policy. One difference, from other litter apps, is that it appears you can track the amount and type of trash you have collected via tags without the need to snap photos of the litter. This helps provide statistics and trends on the litter accumulating within an area, but potentially limits the ability to use the data for AI / machine learning purposes. #CleanSwell also gamifies their app by allowing you to earn badges based on the type and quantity of litter collected. Like Litterati, Clean Swell currently allows you to access historical records of your cleanup efforts, but not the efforts of others; it is not apparent if any of the data can be downloaded. #OceanConservancy is a non-profit organization that has been around for 30 years that has seen over 12 million volunteers collect over 220 million pounds of trash. For questions, comments, or to volunteer please reach out to their general inquires address.
Litter Intelligence: is a coastal, litter based data collection effort that takes place throughout New Zealand, which was launched in 2018 and is led by the charity Sustainable Coastlines. The program follows the United Nations Environment Program / Intergovernmental Oceanographic Commission Guidelines on Survey and Monitoring of Marine Litter, and citizen scientist must first receive basic data collection training before being able to partake in the effort. A breakdown of the training methodology can be found here. #LitterIntelligence is similar to the platforms discussed previously in that a smartphone is used to document each piece of litter collected, however only data from trained individuals can be included in the database, and the data is subject to additional audits for quality assurance. Once collected, the data is hosted on Microsoft Azure,
a cloud computing platform, and then visualized using Microsoft Bi, an analytics platform that allows easy creation of reports and dashboards encompassing the data. To date over 3900 citizen scientists have been trained, 122 monitoring sites established, and 349 coastal litter surveys conducted throughout New Zealand. Litter Intelligence is funded by the Ministry for the Environment’s Waste Minimization Fund, and all data is made free and available to the public. They are seeking global partners interested in instituting the program within their own countries; Expression of Interest (Global) form.
World Waste Platform (WWP): is an effort led by The Lets Do It Foundation with the goal of moving to a zero waste economy by embracing the power of technology. The #WorldWastePlatform leverages data from 8 trash mapping apps, which was compiled by Gray’s Lab. The data breaks the globe apart into three categories (i.e. clean, unclean, and hazardous) and allows for filtering by both country and the app used to collect the data.
Additionally, the WWP partnered with SIFIR Information Technology and Services to train an AI algorithm to detect trash in geolocated images. The training involved collecting ~2000 images, selecting appropriate images to feed into the algorithms, marking the trash in each selected image, iteratively training the AI algorithms, and then testing the algorithms on images of trash they had not previously seen. There were two main takeaways from the AI study. Firstly, the training model leveraged (i.e. Mask R-CNN) allowed for both litter and the background associated with each image to be tagged separately, resulting in much higher accuracy for correctly identifying litter as compared to models in which only litter was tagged. Secondly, the project identified the importance of including images from multiple datasets. Initially, the algorithm was trained on ~600 still images, collected by volunteers at Lets Do It World, and accurately identified trash when presented with similar still imagery it had not previously seen. However, when further testing was conducted using Google Street View Images the algorithm performance declined due to intricacies encountered with images collected while in motion. Ultimately, ~1300 Google Street Images were screened and included as training data for the algorithm, which resulted in increased accuracy of litter detection.
These efforts allowed WWP to detect the exact location of mismanaged trash piles all throughout the world. The Let’s Do It Foundation / WWP is an Estonian non-profit entity, which receives funding from various Estonian governmental bodies; they are currently looking for funding partners to help verify their AI platform. The algorithm code is open to the public and can be found here or on GitHub. The recommended contact for further information is Kristiina Kerge.
Seagull Robotics: was an Indiegogo campaign launched by Andrew Dodd in 2019 with the intent of combining an app named Trashd, similar in concept to the other apps above, and autonomous drones for trash recovery; drones would be trained via AI using images captured via the Trashd app. However, it appears that only three backers supported the initial effort and that further progress and updates were not supplied beyond June 2019; currently the website seagullrobotics.co is down. Nonetheless, the proposal highlights three of the main components needed to re-imagine litter collection: a decentralized method of consolidating a library of reliably tagged litter images, an AI algorithm that can be trained to make sense of the images, and a method to dispatch autonomous resources to cleanup affected areas.
So there you have it, choose your tool of choice and get out there to start cleaning up some litter.
Thanks for stopping by…we’ll see you next time!