Illumination aware raw detection improves objects seen in dark or raw images
ILLRAW: Illumination-Aware RAW Object Detection
Computer Vision and Pattern Recognition
Summary
Detecting objects directly from RAW camera data is tricky because most methods first convert this data into pictures that look nice to humans, losing some useful information. The authors propose a new approach called ILLRAW that skips making pretty images and instead teaches the system to detect objects straight from the raw sensor data. It pays special attention to lighting conditions and how sensor colors behave to get better results. Tests show this method works especially well in low light and challenging exposures.
What this means in practice
- •For camera system engineers: Improve object detection directly from RAW sensor data in cameras, enhancing performance under varied lighting without relying on image conversion.$Commercial implications: Enables camera manufacturers to build smarter detection features that work well in low light and save processing resources.
- •For autonomous vehicle developers: Achieve more reliable object detection from raw sensor inputs especially in poor lighting conditions, aiding safer navigation.
Authors
Xinzhe Wang, Xiao Liang, Xiaoli Zhang, Kang Ma
Abstract
Existing RAW-based detectors typically introduce learnable ISP modules to convert RAW inputs into sRGB-like formats to fit off-the-shelf detectors. However, this RAW-to-sRGB paradigm decouples representation learning from the downstream detection task. Crucially, it forces the network to inherit human-centric visual priors while discarding intrinsic RAW advantages, such as linear sensor responses and high-bit-depth, minimally processed scene cues. In this paper, we propose \textbf{ILLRAW}, an \textbf{ILL}umination-aware end-to-end \textbf{RAW} object detection framework. Instead of reconstructing visually plausible sRGB images, ILLRAW directly learns detection-oriented RAW features through a lightweight feature adapter inserted at the input stage of existing detectors. ILLRAW estimates an illumination mask and introduces channel-sensitive embeddings to model Bayer-channel responses. These features are fused by an illumination attention fusion module. Experiments show clear improvements on the diverse AODRaw benchmark and competitive performance on LOD and PascalRAW, with particularly strong results under dark exposure.