
1 / 5 · Click to enlarge
AI Summary
A self-contained AI vision sensor that runs 20+ models on-device with 6 TOPS of acceleration. It gives Arduino, Raspberry Pi and ESP32 projects real-time object detection and tracking through a touchscreen interface.
Key Highlights
- 6 TOPS AI acceleration for real-time on-device inference
- More than 20 pre-trained models ready to use out of the box
- 2.4-inch IPS touchscreen for model selection and tuning
- Replaceable 2MP GC2093 camera module at 60 FPS
- USB-C, UART and I2C connectivity with major controller boards
- 8GB eMMC storage for models and captured data
Ideal For
- Educators building AI and robotics lessons
- Makers adding vision to Arduino or Raspberry Pi projects
- Students prototyping gesture and face recognition
- Developers deploying custom YOLO models on-device
- Interactive installation designers needing local vision
Buying Tip
Check whether you need the optional Wi-Fi 6 module, since the HUSKYLENS 2 ships without Wi-Fi installed and relies on USB-C, UART or I2C out of the box.
Elecvora expert review
Our verdict
The HUSKYLENS 2 is a well-rounded vision sensor that removes the usual cloud and host-computer barriers from AI projects. Its 6 TOPS of compute, touchscreen control and broad controller support make it a practical choice for education and prototyping.
AI-assisted analysis based on the manufacturer's specifications. Not a hands-on lab test.
Pros
- On-device AI with 6 TOPS keeps latency low and removes cloud dependency
- 20+ built-in models plus custom YOLO support cover many tasks
- Touchscreen makes setup and model switching straightforward
- Works with Arduino, Raspberry Pi, ESP32, UNIHIKER and micro:bit
Cons
- Wi-Fi 6 is optional and not installed by default
- Power draw of up to 3 W requires a supply that can deliver it
Best for
Educators, students and makers adding vision to robotics projects
Not ideal for
Projects that need Wi-Fi out of the box without adding the module
Specifications
| Operating Voltage (VDC) | 3.3 to 5 |
|---|---|
| Computing Power | 6 TOPS |
| Storage Capacity | 8GB eMMC |
| Power Consumption | 1.5 – 3 W |
| Camera Details | GC2093, 2 MP, 1/2.9", 60 FPS |
| Display | 2.4″ IPS Touchscreen (640 × 480) |
| Connectivity | USB-C, UART, I²C, Wi-Fi 6 (optional) |
| Operating Temperature (°C) | -10 °C to +55 °C |
| Compatibility | Arduino, Raspberry Pi, ESP32, UNIHIKER, micro:bit |
| Length (mm) | 70 |
| Width (mm) | 58 |
| Weight (g) | 90 |
| Shipping Weight | 0.18 kg |
| Shipping Dimensions | 11 × 9 × 55 cm |
What's in the box
- 1 x DFRobot HUSKYLENS 2 AI Vision Sensor
- 6 x M3 screws
- 6 x M3 nuts
- 1 x Mounting bracket
- 1 x Heightening bracket
- 1 x Gravity-4P sensor connector cable (30 cm)
- 1 x Dual-plug PH2.0-4P silicone cable (20 cm)
- 1 x Power adapter board
Frequently Asked Questions
Does the HUSKYLENS 2 need an internet connection?
Which boards can I connect it to?
What voltage does it require?
Can I use my own trained models?
What is included with the sensor?
How should I care for the sensor?
Overview
The DFRobot HUSKYLENS 2 is a next-generation AI vision sensor built for makers, educators and developers who want machine vision without a cloud service or a powerful host computer. Inside is a processor with 6 TOPS of AI acceleration, which lets the sensor run detection and recognition models locally and in real time. More than 20 built-in models cover common tasks such as object detection, face recognition, gesture control and object tracking, and you can also deploy your own YOLO-trained models for specialised jobs.
Because the sensor handles the vision workload itself, your Arduino, Raspberry Pi or ESP32 board only has to send commands and read results. The 2.4-inch IPS touchscreen shows the live camera feed and gives you on-screen controls for choosing and adjusting models, so setup does not require a separate monitor.
Key Features
- 6 TOPS of AI compute for fast, on-device inference without cloud dependency.
- 20+ pre-trained models ready to use for detection, recognition, tracking and gestures.
- Custom YOLO model support for tasks the built-in models do not cover.
- Replaceable 2MP GC2093 camera capturing at 60 FPS.
- 2.4-inch IPS touchscreen at 640 x 480 for live view and control.
- 8GB eMMC storage for models and data.
- Flexible connectivity over USB-C, UART and I2C, with optional Wi-Fi 6.
Specifications Explained
6 TOPS of computing power is the key figure here: it describes how many AI operations the sensor can perform per second, which is what allows models to run smoothly on the device rather than on a remote server. The GC2093 camera is a 2MP sensor at 1/2.9 inch, recording at 60 FPS, so the live feed stays smooth during tracking. The 2.4-inch IPS touchscreen at 640 x 480 gives a clear view of what the sensor sees and lets you interact with models directly. 8GB of eMMC storage holds the model library and captured data, and the 3.3 to 5 VDC input suits common controller boards. Power draw of 1.5 to 3 W keeps the sensor within the range of typical small robotics power supplies, and the -10 to +55 °C operating range covers normal indoor and sheltered use.
Compatibility and Requirements
The HUSKYLENS 2 works with Arduino, Raspberry Pi, ESP32, UNIHIKER and micro:bit. It accepts 3.3 to 5 VDC and communicates over USB-C, UART or I2C, so it can slot into existing robotics builds without a dedicated vision computer. Wi-Fi 6 is available as an option but is not installed by default, so plan for a wired connection unless you add the module.
Typical Uses
- Educational robotics projects that need vision without complex setup.
- Interactive installations responding to faces, gestures or objects.
- Prototyping smart devices with on-device detection.
- Custom YOLO model deployment for specific recognition tasks.
- LLM interaction through the built-in MCP service for contextual awareness.
What's in the Box
You receive the HUSKYLENS 2 sensor, six M3 screws, six M3 nuts, a mounting bracket, a heightening bracket, a 30 cm Gravity-4P sensor connector cable, a 20 cm dual-plug PH2.0-4P silicone cable and a power adapter board.




Quick specs
- Operating Voltage (VDC)
- 3.3 to 5
- Computing Power
- 6 TOPS
- Storage Capacity
- 8GB eMMC
- Power Consumption
- 1.5 – 3 W
- Camera Details
- GC2093, 2 MP, 1/2.9", 60 FPS
- Display
- 2.4″ IPS Touchscreen (640 × 480)


















