2DNeuralVision has created a test car with multiple cameras that capture visible and short-wave infrared images simultaneously e.g. in fog, rain and lowlight environments. This achievement is essential for developing reliable, any weather computer vision systems for safer mobility.

Why this matters for Europe

Adverse weather continues to be one of the biggest challenges for automated driving and road safety. Wide-spectrum image sensors are well positioned to provide reliable vision under these adverse conditions. To prove real-world value, wide spectrum imaging needs to be compared apples to apples to other vision solutions.  We need data across VIS and SWIR, captured at the same instant, in various adverse weather conditions. That’s the only way to quantify contrast gains and build trustworthy training data for our Optical Neural Network (ONN):

• Real‑world proof: fog, rain, low sun and nighttime can now be tested side‑by‑side.
• Better decisions: performance gains (or limits) come from data, not guesswork.

By generating directly comparable, timestamp aligned data across different imaging technologies, 2DNeuralVision contributes to:

• Fair, scientifically robust benchmarking of sensor performance.
• Better AI training data for optical neural networks in challenging visibility
• Reduced uncertainty in perception system design for future European mobility solutions

This progress supports the EU’s priorities in safer transport, AI reliability, and low power photonic sensing technologies.

What changed

The new test platform integrates:

• Global timestamps and software triggering to ensure camera synchronisation.
• Controlled sessions (e.g., fog/rain tunnels) are combined with real road drives.
• A unified capture timeline, making annotation, dataset curation and model training significantly faster and more consistent.

What’s next

More scenarios (vulnerable road users at night, low‑contrast “lost cargo”) and simple graphics to show before/after visibility for each camera type.