- New
Product description: AI camera - HUSKYLENS 2 - Kendryte K230 - GC2093 2 MPx - DFRobot SEN0638
The Gravity HUSKYLENS 2 is a next-generation AI camera designed for edge computing applications in robotics, STEAM education, and maker projects. The device is powered by a dual-core Kendryte K230 1.6 GHz processor with a built-in 6 TOPS AI accelerator and 1 GB of RAM, allowing it to run complex neural network models without the need for a cloud connection. Integrated 8 GB eMMC memory allows for storing models and data directly on the device. The sensor is equipped with a 2.4-inch IPS touchscreen display with a 640 x 480 resolution, a removable 2 MPx 60 fps camera module , a MEMS microphone, a 1 W speaker, and LEDs . The HUSKYLENS 2 communicates via USB Type-C and Gravity (I2C/UART) interfaces, while an optional 2.4 GHz WiFi 6 module enables wireless image streaming and remote monitoring of the system.
AI camera - HUSKYLENS 2 - Kendryte K230 - GC2093 2 MPx - DFRobot SEN0638.
You can replace the standard lens yourself with a wide-angle module, a macro lens or an infrared sensor.
Main features of HUSKYLENS 2
- Kendryte K230 1.6GHz + 6 TOPS Processor : Powerful cloud-free AI processing and real-time image analysis
- 20+ built-in AI models : face, object, gesture, pose, OCR, QR code, license plate recognition
- Support for custom YOLO models : Ability to deploy custom neural networks directly on the device
- 2.4'' IPS 640 x 480 px screen : intuitive operation thanks to the touch panel
- Removable 2 MPx 60 fps camera module : compatible with standard, microscope and night vision modules
- MCP Service : context-aware integration with LLM models
- Interfaces : USB Type-C, Gravity 4-pin (I2C / UART), TF slot
- Power supply : 3.3 V to 5 V
HUSKYLENS 2 vs HUSKYLENS 1
AI 2MPx Camera Capabilities
The GC2093 2 MPx 1/2.9" 60 fps sensor ensures smooth, real-time image processing. The camera supports face detection, object recognition, instance segmentation, line tracking, hand gesture analysis, human pose detection, and OCR . The interchangeable optical module design allows the system to be adapted for long-range monitoring, laboratory analysis, or night-time operation. Multi-model operation is possible – running several algorithms in parallel or sequentially.
HUSKYLENS 2 uses the K230 unit (1.6 GHz), offering many times higher computing performance.
Thanks to its broad compatibility with Arduino, Raspberry Pi and ESP32, HUSKYLENS 2 is the perfect tool for developers of interactive systems and STEAM education.
MCP and LLM integration in AI systems
The built-in MCP (Model Context Protocol) service enables the transfer of structured data about detected objects and events to LLM models. Instead of raw images, the system generates contextual information that can be analyzed by linguistic models . This allows robots and intelligent devices to make more informed decisions, respond to specific people or actions , and build advanced human-machine interactions.
Practical applications
HUSKYLENS 2 is suitable for mobile robots, manipulators, STEAM educational systems, and intelligent device prototypes. It can be used with Arduino , Raspberry Pi , ESP32 , micro:bit , and UNIHIKER via I2C or UART interfaces. It is suitable for facial recognition, object tracking, behavior analysis, and interactive projects requiring real-time image processing.
Contents of the set
- 1x HUSKYLENS 2
- 6x M3 screws
- 6x M3 nuts
- 1x mounting bracket
- 1x Elevator handle
- 1x Gravity Cable 4-pin 30cm
- 1x Silicone cable PH2.0-4P with two plugs 20 cm
- 1x Power board
Elements included in the set.
| Technical specifications and comparison HUSKYLENS 1 vs HUSKYLENS 2 | ||
|---|---|---|
| Parameter | HUSKYLENS 1 (SEN0305) | HUSKYLENS 2 (SEN0638) |
| Processor | Kendryte K210 400 MHz | Kendryte K230 Dual-Core 1.6 GHz |
| AI computing power | Basic Edge AI | 6 TOPS |
| RAM | Built into the K210 system | 1 GB LPDDR4 |
| Internal memory | No eMMC memory | 8GB eMMC |
| Image sensor | OV2640 / GC0328, 2MPx | GC2093, 2 MP, 1/2.9", 60 fps |
| Screen | 2'' IPS, 320 x 240 px + knob | 2.4'' IPS touchscreen, 640 x 480 px |
| Built-in models/algorithms | Facial recognition Object tracking Object recognition Line Tracing Color recognition Tag Recognition Classification of objects | Face detection Facial recognition Facial feature detection Object recognition Object tracking Color recognition Classification of objects Self-learning classifier Instance segmentation Hand detection Hand keypoint detection Hand gesture recognition Human silhouette detection Body Keypoint Detection Body pose recognition License plate recognition Text recognition (OCR) Line Tracing Facial emotion recognition Gaze direction detection Face orientation detection Tag Recognition QR code recognition Barcode recognition Fall detection |
| Own models | Simple classification learning | Implementing YOLO models (custom neural networks) |
| LLM Integration | Lack | MCP (Model Context Protocol) |
| Built-in applications | No dedicated system applications | Camera, Video Recorder, Real-Time Video Transmission, MCP Service, Model Deployment |
| Microphone | Lack | MEMS |
| Loudspeaker | Lack | 1 In |
| Illuminating diodes | 1 x LED | 2 x LED |
| RGB indicator | Lack | 1 x RGB LED |
| Interfaces | UART, I2C | USB Type C x1, Gravity 4-pin (I2C / UART) x1, TF slot x1 |
| Wireless connectivity | Lack | 2.4 GHz WiFi 6 (optional module) |
| Expandable memory | Lack | TF slot |
| Operating voltage | from 3.3 V to 5 V | from 3.3 V to 5 V |
| Energy consumption | 320 mA at 3.3 V / 230 mA at 5 V | from 1.5 W to 3 W |
| Dimensions | 52 x 44.5 mm | 70 x 58 x 19 mm |
Useful links |
