A series of projects which i self initiated, formed teams, led, designed, developed on how AI can help WildLife conservation. Started in 2017, with active development till 2020. Since then i have been mentoring startups through Nasscom Ai Challenge, and also advising Telangana Govt, Microsoft Sales, Azure team in this domain.
Project 1 : Individual species detection in camera trap images
Abstract:
India is home to the highest diversity of wild cat species, and regular censuses are conducted to monitor their populations. The Wildlife Institute of India (WII) surveys endangered species like tigers, leopards, and snow leopards using camera traps in remote forest areas. These cameras often capture "blank" images triggered by wind. The lack of awareness and government support leads to roadkill, retaliatory killings, and other threats to these animals.
This project aims to leverage AI technology to address these conservation challenges. By adopting a user-centric design process, we focused on real needs rather than applying technology for technology's sake. Growing up in India's forests and wildlife, I’ve always been passionate about conservation. Two years ago, I began volunteering with the Eastern Ghats Wildlife Society, further strengthening my commitment to this cause.
User Studies:
A field trip to Krishna Sanctuary was conducted with the Eastern Ghats Wildlife Society (EGWS) to search for the endangered, elusive fishing cat. Due to its nocturnal behavior, evidence of the species is scarce. Camera traps were placed in mangrove forests after a 3-hour boat journey and left for 15 days to gather data. However, many of the images collected were blank, false positives, or of domestic animals. In some cases, the traps were washed away by the sea, and when valid images were captured, the animals had often moved too far to track.
Typically, 100-200 camera traps are installed for such surveys, and much time is spent sorting useful images from irrelevant ones. Once a sufficient number of images are obtained, researchers manually identify individual animals based on unique patterns, a method used for patterned species like tigers and leopards.
The lack of awareness among locals, along with insufficient evidence and surveys, has led to roadkills, retaliatory killings, and hunting for bushmeat, all contributing to the decline of these species. Additionally, the government does not prioritize these animals as they fall outside the "charismatic megafauna" category.
Hence problems identified:
Personas:
Two key stakeholders in this effort
- Wildlife biologist
- Local villager who lives nearby forests with sufficient internet connectivity in case an animal stray out – which is very often
Problem statement
Brainstorming: To counter the above problems we came up with a number of ideas such as
- Enhancing community participation through apps like wildlife tracker – using paw marks sign surveys to know about animals and track them
- Better maps for data visualization of terrain
- Sending signals through camera traps once an animal is detected
- A dashboard for tracking animals through wearable sensors, cameras
- Awareness campaign, social media
AI was seen as a key solution for wildlife conservation, leveraging its ability to "see, predict, and act." By automatically identifying animals through images of paw prints, tracks, and scat, we aim to educate the public and streamline the process of identifying animals in camera trap data. Interactive 3D maps will be used to present the story of wild cats in an engaging, informative way.
Prototyping and Test
The project shifted focus to larger cats, such as tigers and leopards, due to better data availability and new stakeholders like the Wildlife Institute of India and Corbett Tiger Reserve. We evaluated ideas based on factors like need, technical complexity, relevance, and impact, with AI drawing significant interest due to its novelty, efficiency, and government backing, which could secure funding. The selected ideas for development include:
- AI for Animal Detection: AI will identify whether an animal is present in camera trap images and specify the species.
- Patterned Species Identification: AI will identify individual animals based on unique patterns (e.g., stripes or spots).
- Tourist Engagement App: Tourists will use a mobile app to learn about animals by uploading pictures, fostering greater public awareness.
- Drone Surveillance: Drones equipped with IR cameras will monitor for poaching and encroachment in protected areas.
AI’s role also attracts interest from local technical schools. The focus shifted to tigers and leopards due to their endangered status and a more comprehensive dataset. For wildlife researchers, a progressive web app was chosen for desktop use, while a mobile app for tourists and villagers was developed. Both apps feature a responsive design with different interfaces for researchers and locals. To boost engagement, a leaderboard was added, alongside data visualization tools for researchers to track biodiversity.
Customvision was used to identify the animals.
Project 2: Keeping pangolins offline and in the wild:
Independent initiative : Telangana Ai Mission, Microsoft Azure
Project 4: Poacher detection with fixed winged drones:
SPOT Poachers in action : Augmenting Conservation Drones with Automatic Detection in Near Real Time. Detection in thermal infrared imagery captured aboard a UAVRunning in near real time with a laptop
For the drone’s solution much of the problem I focused was on getting the right kind of drones – long range, fixed wing drones to Corbett Tiger Reserve. Parallelly in collaboration with University of southern California, Microsoft Research, efforts were made to get the backend running and deployed at CTR as a pilot programme.
.jpg)