AI4COYOTES: Preventing coyote-human conflicts using Artificial Intelligence

Developed a Deep Neural Network AI system to automatically detect coyotes near human habitations and notify residents using mobile alerts.

Photo of Vivek Bharati
3 11

Written by

Please confirm you meet the following criteria

  • We have submitted the supplemental form linked in the description above
  • We are aged between 14 - 20 as of February 11, 2021
  • We live in the United States or its territories (Puerto Rico, U.S. Virgin Islands, Guam, Northern Mariana Islands, and American Samoa)
  • We are not employed by, or directly related (parents or siblings) to a current General Motors (GM) or Ashoka employee
  • We have been working on this project for at least three months
  • We consent to Ashoka and/or GM featuring our work on their website, social media, and in other materials regarding this Challenge using the information in our application
  • We confirm we have the rights to use and share any content uploaded on this entry form

Website or social media url(s) (optional):

ai4coyotes.org

Date You Started Your Project

6/24/2019

Project Stage: Select the description below that best applies to your approach.

  • Established (successfully passed early phases, have a plan for the future)

1. The Problem: What problem are you helping to solve?

Coyotes are beautiful and intelligent animals that inhabit practically the whole of North America. They are known to be devoted caregivers who typically live with a single companion all their adult life and lovingly raise their young. Yet, they are much maligned and harshly persecuted in areas where their habitats are near human settlements. Most conflicts between coyotes and humans occur when humans are surprised by the presence of the coyote.

2. Your Solution: How are you planning to solve this problem? Share your specific approach.

Our premise is that most coyote-human conflicts can be avoided if humans who could potentially be affected can be forewarned about the presence of the coyotes in time so that they can take preventive measures. For example, if humans could know that a coyote is headed towards their homes, they can bring in their small pets and secure their lifestock. The alerts must be accurate and timely so that such protective measures can be taken. Avoiding the loss of valuable domesticated animals takes away the motivation of humans to attempt to eliminate coyote populations. Likewise, if coyotes do not find easy prey, they are not likely to have the strong motivation to come into human habitations. People carry mobiles all the time. So mobile phones are the best way to send alerts to the people who may be affected. Deep Neural Networks have made it possible to identify specific animals in images. We plan to use a system with a two stage Deep Neural Network, we call CoyoteNet, that use images from home surveillance cameras or low cost cameras placed near typical coyote paths to detect presence of coyotes and alert only those living near the specific location where such detection occurs.

3. Please tell us how you are using science, technology, engineering or math to address your environmental challenge.

The key to protecting the coyote is to detect its presence using technology and then quickly alert neighbors in the area. Vivek had learned about Deep Neural Networks (DNNs), an AI technology, almost a year and a half back. He realized that a 2 stage DNN - object detection Convolutional Neural Network (CNN) followed by a classification CNN - can be created to automatically detect a coyote from an image, if trained with large number of images of coyotes. This was our inspiration. We set about creating the most optimal configuration of DNNs that can do this with minimal power and memory requirements. We had to create a custom DNN model and then experiment with various configurations to arrive at the optimal configuration. The model was trained using a large number of coyote images from the Caltech wildlife image data set. We then designed a system around the DNN so that location of the coyote can be noted and then mobile alerts sent to people in just the required neighboring areas. The mobile alerts are sent using a cloud based notification system. The system we have created is effective and can be made as part of most home surveillance cameras or in specialized low cost devices.

4. Personal Journey: What’s the story behind why you decided to start this project?

Our neighborhood is located right next to the Rancho San Antonio Preserve, a 4,277 acre public recreational area in silicon valley and home to a variety of wildlife including coyotes. Every now and then we would get an email in our neighborhood group that a coyote has been sighted. This would immediately be followed by a flurry of responses asking everyone to bring their pets indoors and often painting the picture of a vicious animal on the loose. It made us curious to know more about them. When we researched about the coyotes, we realized that they are very intelligent and social animals that specialize in caring for their families, often with a single partner over their lives. We then learned about coyote hunts that wreak havoc on the coyote population. We realized that the way to stop this is to prevent the coyotes from surprising humans. We knew deep learning technology could help.

5. Video (Keep it simple, your phone on selfie-mode is great): Please upload a 1-minute video to YouTube that answers the following “I am stepping up to be a Changemaker because...”

6. Please highlight the key activities you have carried out to bring your project to life.

The first activity was to validate the hypothesis that Deep Neural Networks (DNNs) can in fact detect coyotes. We began with a starting model architecture and used the Caltech training data to train this model. We found that it is in fact possible to detect coyotes from an image using DNNs. Then we set about optimizing the model by varying the model's parameters. This yielded a configuration with low memory and processing consumption. We then added the mobile notifications from Amazon cloud.

7. The X Factor: What is different about your project compared to other programs or solutions already out there?

This project's X Factor is the use of cutting-edge Artificial Intelligence technology to automatically detect a coyote, along with mobile technology to alert just the right set of people who need to take precautions when the coyote is sighted. By contrast, neighborhood groups rely on people watching out for coyotes manually and using mechanisms such as emails to alert other neighbors, often contributing to more animosity towards coyotes. At times, coyote hunts are organized to kill coyotes which actually compounds the problem by unbalancing coyote population. Our technology is also affordable.

8. Impact: In the last three months, please detail the impact your project has made.

We have been able to experiment with a number of possible neural network architectures and arrive at the best architecture and the specific configuration of this architecture for detecting coyotes from images typically available from surveillance cameras or specialized cameras. We have also been able to identify the source of training data for this neural network. As a result of these efforts in the last three months, we now have a pre-trained neural network model that can detect coyotes from an image and the system to send notifications to mobiles upon such detection. We now have embarked upon the task of connecting with organizations around the country who are championing the cause of the coyote with the objective of leveraging their connections to deploy this technology via companies that make surveillance cameras. We plan to license our solution software free of cost to any company.

9. What’s Next: What are your ideas for taking your project to the next level?

Now that we have a robust solution, we will be focusing on getting the buy-in of the community ecosystem to deploy it on a large scale across North America. For this, we need strong motivated evangelizers who are aligned with the cause of championing the coyote. The first partnerships we plan are with national non-profit organizations that on a mission to protect native predators in America. Our technology solution will be a nice complement to their advocacy. We have reached out to two organizations that are on a mission to protect the coyotes. The next step would be to encourage makers of standalone and surveillance cameras to embed CoyoteNet in their products. We plan to offer our technology free of cost for embedding in their products.

10. Please share how you have influenced other young people to get involved in your project and/or care about environmental sustainability.

We have been able to influence peers by a combination of empathy and persuasion to remove fear of the coyote and to instill a sense of urgency. When Vivek reached out to a friend asking for support, her first reaction was 'Why the Coyote? They are dangerous'. He explained about the social nature of coyotes and their caregiving nature such as their companionship for life. Then he talked to her about how such misunderstanding has led to indiscriminate hunting and how to prevent it with technology.

11. How would you partner with other changemakers to make a difference?

We will need the help of other changemakers to create awareness about the true nature of the coyote and mobilize the society towards adopting a technological solution to protect it. We will ask them to write to their friends and family to first change the negative perception about the coyote. Young changemakers are more receptive to the use of technology. Once they are convinced, we will aim to convince law makers to mandate use of technology solutions before any coyote hunts can be approved.

12. How would you engage others who have never heard about your project to get their buy-in?

Buy-in will be obtained in two steps. First is to remove the negative perceptions about the coyote. From narrating native American Ohlone lores about the coyote to talking about their unique social nature, we have to convince people that coyotes are misunderstood. The next step would be to convince people that a technology solution such as ours has just now become possible due to advancements in artificial intelligence technologies. Demonstrations of our technology solution in action will help.

13. Finances: If applicable, have you mobilized any of the following resources so far?

  • Friend support
  • Family support
  • Mentors/advisors

14. Which of the following types of expertise would be most useful for you? You’ll be able to select only one option.

  • Marketing Strategy

Are you employed, or directly related (grand-parents, parents, sibling) to a GM or Ashoka employee?

  • No

How did you hear about this challenge?

  • Search engine

Complete this evaluation to see the results from the rest of the community

The community is currently evaluating this idea.

Attachments (1)

AI4Coyotes Technical Paper.pdf

Technical description of our CoyoteNet solution.

3 comments

Join the conversation:

Comment
Spam
Photo of Sophia Pryor
Team

I love this idea! It's amazing that you're all working to change community behavior through deepening peoples understanding of our natural environment. I think the two pronged approach of outreach/education on coyotes and working to keep people safe by alerting them when coyotes are nearby is a great way to tackle this issue. Have you already reached out to coyote advocacy organizations already with this idea? I think connecting with them early in the project will be a great way to get feedback from the experts and make a better product with the most impact! Great work, keep it up!

View all comments