The OAK-T (OAK Thermal) is our first camera with a thermal sensor and a color image sensor.

When fusing thermal frames, color frames, and AI models, you can create a wide variety of applications in different fields, such as:

• Agriculture: Health crop, detect irrigation issues, detect pests, livestock health monitoring, etc.
• Industrial: Detect overheating, detect leaks, detect fires, etc.
• Security: Detect intruders (night vision), detect fires, etc.
• Transportation: Detect pedestrians, vehicles, bicycles for analytics and traffic optimization.

Please note that due to legal restrictions, we can only ship OAK Thermal cameras to member countries, Australia, Israel, Japan and South Korea.

<!–--split--–>

Description

OAK-T leverages our to make an overall compact design. The use of the SoM reduces the design's scale, making it easier to mount or fit anywhere seamlessly. The design will also be open-source when we verify the hardware design.

The purpose of the device is to be used for manufacturing, warehouse, energy efficiency, and safety monitoring.

PoE connection allows cameras to communicate to other computers via networking protocols (, UDP, , , etc.). As our PoE models feature device flash memory, you can also run your programs in . OAK PoE cameras are housed in an rugged, IP-rated enclosure, which is required for industrial and outdoor use.

We suggest following as the first steps after installation.

RVC2 architecture on-board features:

• 4 TOPS of processing power (1.4 TOPS for AI - )
• Run any AI model, even custom architectured/built ones (models need to )
• Encoding: H.264, H.265, MJPEG - 4K/30FPS, 1080P/60FPS
• Computer vision: warp (undistortion), resize, crop via node, edge detection, feature tracking. You can also run custom CV functions
• Object tracking: 2D tracking with ObjectTracker node

• Connectivity: M12 PoE connector and M8 connector for IO connectivity
• Thermal sensitivity: 50mK @ 25°C and 25Hz
• Measurement range: -15°C - 150°C
• Thermal sensor resolution: 256x192
• Thermal Accuracy: ±2°C or ±2% (whichever is larger)
• Sensor type: Uncooled VOx detector
• Spectral range: 8-14µm

• Power consumption: up to 6.5W

• Industrial grade aluminum housing with Front Gorilla Glass
• 1/4" tripod mount and 75mm M4 Vesa mount

• Dimension: 80x52x46mm
• Weight: 297g (with enclosure)

Product Diagram

Technical Specifications

Power-over-Ethernet (PoE)

The base board offers full 802.3af, Class 3 PoE compliance with 1000BASE-T speeds (1Gbps). PoE injector/switch is required to power the device. If need an injector, you can buy it here.

Sensors specifications

Power consumption

Most of the power is consumed by the RVC2, so the power consumption mostly depends on the workload of the VPU:

• Base consumption + camera streaming: 2.5W - 3W
• PoE circuitry: 0.5W
• Thermal sensor: up to 1W
• AI subsystem consumption: Up to 1W
• Stereo depth pipeline subsystem: Up to 0.5W
• Video Encoder subsystem: Up to 0.5W

So the total power consumption can be up to ~6.5W if you are using all the features at 100% at the same time. To reduce the power consumption, you can reduce FPS of the whole pipeline - that way, subsystems won’t be utilized at 100% and will consume less power.

Thermal perception

Thermal cameras detect infrared radiation, which is emitted by all objects based on their temperature. We use a special IR-capable lens to focus the infrared light emitted by all of the objects/scene in view, which is then scanned by an array of infrared-detector elements (“pixels”) in the camera.

Thermal perception range

Thermal perception range is similar to color camera’s range, it mostly depends on the object’s size and temperature. For example, with you can easily detect a large building even from 1km away, but you won’t be able to detect a small animal from that distance. The following table shows the approximate thermal perception range for different object sizes:

The DRI (Detection, Recognition, Identification) ranges are based on the Johnson's criteria, which is a standard for optical systems. Specifications are as follows:

• Detection: The ability to notice that something is there, but without the ability to identify it
• Recognition: Being able to classify the type of object (like distinguishing a car from a truck)
• Identification: The capability to specify the exact nature of the object (like determining the make and model of a car).

Thermal accuracy over distance

Thermal accuracy decreases as the distance increases. To minimize this effect, we perform a 2-point calibration during factory setup. However, some deviation in measurements remains, influenced by the distance and the actual temperature.

M12 PoE connector is used for power and GBit data transfer. Additionally M8 connector for IO connectivity; USB2 (D+, D-), camera IOs (FSIN, STROBE), power, and one auxiliary GPIO. Cap for M8 is included when connector is not used to ensure water resistance. More information can be found here.

Rotate the locking ring (not the connector body) clockwise (CW) with <0.8 Nm torque to securely connect the cable.

Integrated IMU

This OAK camera has an integrated BMI270, a 6-axis IMU (Inertial Measurement Unit). See IMU node for the API details on how to use it.

Operating characteristics

This camera is shipped with IP67 rated enclosure. More details can be found here.

Temperature

The ambient operating temperature of RVC2 based devices is between -20°C and 50°C when fully utilizing the VPU. To find out more, see our operating temperature range documentation.

Code Examples

• Thermal camera streaming
  
• RGB-Thermal stream alignment - with user-defined depth plane 
• Thermal people detection - using on-device neural network inferencing

Package Includes

Package contains:

• OAK-T camera
• M12 plug to RJ45 cable, 2m
• Cleaning cloth

Note: PoE injector / switch is required to power the device and it is not included in the packaging. If you need an injector, you can buy it here.

Files

1. 3D Models (Downloads STEP file)