Enhanced Wildlife Monitoring
1. Accurate species identification
2. Real-time population tracking
3. Early detection of endangered species and poaching
Data-Driven Conservation
1. Informed decision-making
2. Comprehensive data analysis
3. Understanding animal behavior and ecosystem health
Educational Advancement
1. Enriched content for schools/universities
2. Hands-on learning tools
3. Increased awareness and engagement
Tourism Enhancement
1. Enhanced visitor experiences
2. Interactive, educational tours
3. Increased revenue for conservation
While using ResNet50 for wildlife image classification with PyTorch transfer learning, we encountered a few challenges. One hurdle was likely the class imbalance in our dataset. Wildlife images might not be evenly distributed across all species, potentially leading the model to favor frequently occurring animals. Additionally, background clutter in wildlife photos could confuse the model, making it difficult to distinguish between the target animal and its surroundings. Finally, ResNet50's complexity might have required more training data for optimal performance on our specific wildlife classification task.
Although, we overcame in by preprocessing data suitably and altering the training parameters in order to get the best accuracy.
Another challenge was on integrating the model with the website using flask.
Tracks Applied (1)
Discussion