Shreena Lathiya completed Machine Learning for Automotive Safety

Built a perception-validation dossier that correctly defined detection matching with IoU, computed aggregate precision/recall and VRU false-negative rate from scored detections, and exposed severe recall collapse in night and occluded VRU slices. The work shows careful metric discipline and a safety-first focus on vulnerable road users rather than relying on aggregate accuracy.

• Leakage-Free Detection Definition
• Honest Metric Computation
• Per-Condition Safety Triage
• Safety-Critical Metric Prioritization