• Created an assistive model for visually impaired people to safely cross roads, using Time Distributed LSTM-CNN Hybrid Model and ConvNext architecture.
• Conducted extensive research under Dr. Kinjal Joshi, focusing on refining various deep learning models to enhance real-time object detection.
• Meticulously fine-tuned models and optimized datasets to achieve an F1-score of 88.8%, surpassing previous results by 2% while also reducing latency.
• Demonstrated capability to push the boundaries of computer vision technology, particularly in challenging environments like Indian roads.

Technologies: PyTorch • TensorFlow • Computer Vision • LSTM • CNN • ConvNext