NTT innovation monitors video quality from autonomous vehicle for object detection

Press Release, 10 November 2025

NTT, Inc. has unveiled a new model called Parametric object-recognition-ratio-estimation, designed to automatically determine whether remote-monitoring video streams from autonomous vehicles meet the clarity threshold needed for safe object detection in real time. This innovation, which monitors video quality and alerts operators when it falls below safe levels, has now been adopted by ITU‑T SG12 as the international standard Recommendation P.1199.

The drive behind the technology stems from the rising complexity of advanced autonomous driving systems, especially at Level-4 where human operators remotely oversee vehicles. While sensors and detection algorithms have improved dramatically, one missing link has been a reliable metric for when video quality is simply too poor to safely supervise a vehicle. NTT’s model fills that gap: it analyses incoming vehicle video and related data, estimates the likelihood that suddenly appearing objects would be detected, and triggers alerts if the chance falls below safe thresholds.

Once embedded into remote-monitoring systems, this model promises to bolster the safety of autonomous-vehicle supervision and ensure that operators intervene in a timely way when visibility degrades. The ITU-T’s recognition of the model not only underscores its technical merit but also helps pave the way for broader industry deployment—giving OEMs, operators and regulators a common benchmark for video-quality assurance.

As autonomous mobility expands, having dependable tools like this will be key to maintaining trust, safety and regulatory alignment across different markets worldwide.

Compiled using AI