Inspiration
Our fascination with the prehistoric world, combined with the latest advancements in AI, inspired us to develop DinoClassify. We aspired to create a tool that could go beyond the limited fossil record to visualize the rich tapestry of life that once thrived on our planet, connecting the dots between the known and the unknown in the dinosaur lineage.
What it does
DinoClassify harnesses machine learning to identify and classify images of dinosaurs with remarkable accuracy. Utilizing a refined dataset and cutting-edge neural networks, it offers a glimpse into how various dinosaur species might have appeared, by analyzing features and patterns within the images.
How we built it
We built DinoClassify using a ResNet-18 neural network model, training it on an augmented dataset of dinosaur images. Through meticulous fine-tuning and rigorous validation processes, we optimized the model to recognize and predict with high precision.
Challenges we ran into
Throughout the project, we faced challenges including sourcing quality data, balancing class representation, and optimizing the model to understand the complex features of these ancient creatures.
Accomplishments that we're proud of
We are proud to have achieved a classification accuracy of 90%. But beyond numbers, we are thrilled at having crafted a tool that can educate and inspire, bringing the ancient world of dinosaurs closer to enthusiasts everywhere.
What we learned
The project was a profound learning experience, deepening our understanding of neural networks, data augmentation, and the intricate details required for accurate image classification.
What's next for DinoClassify
Moving forward, we aim to expand DinoClassify with more species, integrate user feedback, and explore the use of our model in educational settings to make learning about paleontology more interactive and engaging.
Log in or sign up for Devpost to join the conversation.