Abstract and Workplan
Abstract and Workplan
Submitted by: -
Nishanth Samson I (23MCA0202), Nasikethan R (23MCA0235), Ram Dhighash N.G. (23MCA0328)
This research paper introduces an advanced visual inference system tailored to assist the
visually impaired in navigating complex environments, incorporating real-time danger
detection, and warning mechanisms. The system seamlessly integrates YOLOv8-based object
detection with Gemini language models to achieve precise object recognition and
environmental description. Our framework, built on cutting-edge technologies, enhances scene
understanding and can also assist the visually impaired by detecting potential dangers such as
oncoming traffic or approaching trains. The YOLOv8 component ensures swift and accurate
identification of objects, while the language model processes detected elements, generating
informative summaries. The extended functionality incorporates a sophisticated danger
detection module, providing timely alerts to users in the face of potential threats. This
comprehensive solution not only empowers the visually impaired with unprecedented spatial
awareness but also adds a critical safety dimension to their autonomous navigation. Evaluation
results are expected to showcase the system's efficacy in detecting and alerting users to
potential dangers, reinforcing its role as a transformative tool for enhancing accessibility and
safety in dynamic surroundings.
Workplan
Problem Identification
Literature Survey
Algorithm Study
Implementation
Analysis
Comparison
Final Presentation