FREE Teledyne FLIR Thermal Dataset for Algorithm Training

This free Teledyne FLIR ADAS Dataset provides fully annotated thermal and visible spectrum frames for the development of object detection systems using convolutional neural networks (CNNs). This data was constructed to encourage research on visible + thermal sensor fusion algorithms ("RGBT") and to empower the automotive community to create safer and more efficient ADAS and driverless vehicle systems.



Why Use Teledyne FLIR Thermal Sensing for ADAS?

The ability to sense thermal infrared radiation, or heat, provides both complementary and distinct advantages to existing sensor technologies such as visible cameras, Lidar, and radar systems.  The Teledyne FLIR thermal sensors can detect and classify in challenging conditions including total darkness, most fog, smoke, inclement weather, and glare.  When combined with visible light data and distance scanning data from Lidar and radar, thermal data paired with machine learning creates a more comprehensive detection and classification system.

Dataset Specifications

Content A total of 26,442 fully annotated frames with 520,000 bounding box annotations across 15 different object categories.
Images 9,711 thermal and 9,233 RGB training/validation images with a suggested training/validation split. Includes 16-bit pre-AGC frames.
Videos 7,498 total video frames recorded at 24Hz. 1:1 match between thermal and visible frames. Includes 16-bit pre-AGC frames.
Frame Annotation Label Totals Over 375,000 annotations in the thermal and visible spectrum.

Video Annotation Label Totals

Over 145,000 annotations in the thermal and visible spectrum

Label Categories
  1. Person
  2. Bike
  3. Car
  4. Motorcycle
  5. Bus
  6. Train
  7. Truck
  8. Traffic light
  9. Fire Hydrant
  10. Street Sign
  11. Dog
  12. Skateboard
  13. Stroller
  14. Scooter
  15. Other Vehicle
Thermal Camera Specifications

Teledyne FLIR Tau 2 640x512, 13mm f/1.0 (HFOV 45°, VFOV 37°)

Visible Camera Specifications Teledyne FLIR Blackfly S BFS-U3-51S5C (IMX250) camera and a 52.8° HFOV Edmund Optics lens
  • Dataset File Format
Thermal - 14-bit TIFF (no AGC)
Thermal 8-bit JPEG (AGC applied)
RGB - 8-bit JPEG
MSCOCO formatted annotations (JSON)
Conservator formatted annotations (JSON)

Related Products

PathFindIR™ II

Driver Vision Enhancement System for Today’s Vehicles


Thermal Vision for ADAS and AV Developers