Detect Wildlife Before It Detects You

• 

  Wildvision in Action

Inspiration

As human-wildlife conflicts grow due to urban expansion, we wanted to use AI to help detect animals early, supporting both safety and conservation.

What it does

Detect Wildlife Before It Detects You uses a pre-trained deep learning model to identify animals in images with high accuracy. The user uploads an image, and the system quickly returns the most likely animal species along with its confidence score. It works through a clean, web-based interface using Gradio, making it usable by anyone with internet access.

How we built it

We used Python as the core language. The frontend was built using Gradio to provide an interactive image upload and output experience. The animal detection model is based on Microsoft’s ResNet-50, accessed via Hugging Face Transformers. We used torch, transformers, and Pillow for backend AI logic and image processing. The project runs locally and can easily be deployed on a web server or cloud platform for wider access.

Challenges we ran into

Setting up the correct environment and handling dependency conflicts (especially with torch, transformers, and PIL) Preprocessing the image data correctly for the model Ensuring the model output was relevant, understandable, and user-friendly Understanding Gradio’s callback and input handling mechanisms

Accomplishments that we're proud of

Successfully integrated a powerful pre-trained AI model into a real-world use case Built a clean, minimal user interface that requires no technical expertise to use Delivered fast and accurate predictions even on a local setup Learned how to use open-source tools for social and environmental impact

What we learned

The power of transfer learning and pre-trained models in rapidly developing AI solutions How to work with Gradio to quickly build UI/UX for machine learning applications Image handling and processing for inference using Python Packaging, running, and troubleshooting Python projects effectively

What's next for Detect Wildlife Before It Detects You

Real-time detection from camera streams (e.g., wildlife cameras, surveillance feeds) Species insights: Provide conservation data, facts, and behavior information along with predictions Localization: Detect multiple animals and their positions in the image using object detection Mobile App: Build a cross-platform mobile app using React Native or Flutter Custom Model Training: Fine-tune on wildlife-specific datasets for even better accuracy

Built With

• gradio
• opensource
• python