Fenglong Song

Papers

Total Citations

H-Index

About

Fenglong Song is a computer vision researcher whose work focuses on advancing object detection under challenging real-world conditions, particularly through the use of RAW sensor data. His most notable contribution is the paper "Toward RAW Object Detection: A New Benchmark and A New Model" (2023), which has garnered 28 citations. This work addresses a critical gap in computer vision: the need for high dynamic range (HDR) data to enable robust object detection in applications like robotics and autonomous driving, where lighting conditions vary dramatically—from strong glare to deep shadows. Song’s research demonstrates that performing detection directly on RAW sensor data, rather than processed images, preserves richer visual information, leading to improved performance in extreme environments. By introducing a new benchmark and model tailored for RAW input, he has paved the way for more reliable perception systems. His work is particularly impactful for engineers and researchers developing safety-critical autonomous systems, offering a practical pathway to handle the unpredictable lighting that often confounds traditional algorithms. Song’s contributions highlight the importance of sensor-aware processing in pushing the boundaries of computer vision.