Datasets | Models | Platforms | Benchmarks |
Welcome to Diesimo Lab your go-to Open Research Lab for Edge Intelligence and Efficiency.
Find out practical resources and recipes to streamline your Edge AI development and experiments, from datasets, models, benchmarks tested on real-world platforms with industry standards in mind.
- Real-world custom-datasets: HF Collections (Coming Soon)
- Model Garden: Hugging Face Collections (Coming soon)
- Demos: Hugging Face Spaces (Coming soon)
In the meantime, find the best model suitable for your application requirements here: Edge AI Model Zoos.
Arduino Nano 33 BLE
| Device | Arduino Nano 33 BLE Sense |
|---|---|
| Processor | 1-Core @64MHz nRF52840, ARM Cortex-M4 32-bit (FPU) |
| Accelerator | None |
| Memory | 256KB RAM |
| Storage | 1MB Flash |
| Power Consumption | Ultra-low power (mW) |
| Connectivity | BLE, NFC |
| Camera/Display Support | None |
| Sensors | 9-axis IMU, temp, humidity, pressure, microphone, gesture, light |
| I/O Pins | 14 digital, 8 analog, 12 PWM |
| Dimensions | 45mm x 18mm |
| Use Case | TinyML, IoT, wearables |
Full Tech specs: (Source)
Raspberry Pi Pico WH
| Device | Raspberry Pi Pico WH |
|---|---|
| Processor | 2-Core @133MHz RP2040, ARM Cortex-M0+ 32-bit (no FPU) |
| Accelerator | None (8x PIO state machines for custom I/O) |
| Memory | 264KB on-chip SRAM |
| Storage | 2MB on-board QSPI Flash |
| Power Consumption | 1.8-5.5V DC input (micro-USB/VSYS), sleep + dormant modes |
| Connectivity | 2.4GHz 802.11n WiFi (WPA3), Bluetooth 5.2 (BLE + Classic), USB 1.1 device/host |
| Camera/Display Support | None onboard (via SPI/I2C modules, e.g. Arducam) |
| Sensors | On-chip temperature sensor |
| I/O Pins | 26 GPIO (3x 12-bit ADC), 2x SPI, 2x I2C, 2x UART, 16x PWM; pre-soldered 2x20 headers + 3-pin debug |
| Dimensions | 51mm x 21mm |
| Use Case | TinyML, IoT, low-power wireless sensing |
Full Tech specs: (Source)
| Device | Raspberry Pi |
|---|---|
| Processor | 4-Core @1.4GHz BCM2837B0, Cortex-A53 64-bit SoC |
| Accelerator | None |
| Memory | 1GB LPDDR2 SDRAM |
| Storage | SD card |
| Power Consumption | 4W (5V, 2.5A ≈ 12.5W, varies by load) |
| Connectivity | WiFi, BLE, Ethernet, USB |
| Camera/Display Support | 8MP (up to 3840 x 2160) |
| Sensors | None |
| I/O Pins | Extended 40-pin GPIO header |
| Dimensions | Not specified |
| Use Case | Lightweight IoT, smart home, education |
Full Tech specs: (Source)
Qualcomm Arduino Uno Q
| Device | Qualcomm Arduino UNO Q |
|---|---|
| Processor | 4-Core @2.0GHz Qualcomm Dragonwing QRB2210, ARM Cortex-A53 64-bit + STM32U585 MCU (1-Core @160MHz ARM Cortex-M33, FPU) |
| Accelerator | Adreno 702 GPU @845MHz (OpenCL), 2x ISP (13MP + 13MP or 25MP @30fps) |
| Memory | 2GB or 4GB LPDDR4X (MCU: 786KB SRAM) |
| Storage | 16GB or 32GB eMMC (MCU: 2MB flash) |
| Power Consumption | 5V @3A via USB-C (PD 5V/3A profile) or 7-24V DC input (VIN) |
| Connectivity | WiFi 5 802.11a/b/g/n/ac dual-band (2.4/5GHz), Bluetooth 5.1, USB-C 3.1 (host/device role switching) |
| Camera/Display Support | 4-lane MIPI-CSI-2 camera, 4-lane MIPI-DSI display, 1080p video out via USB-C (DisplayPort Alt-Mode), 8x13 blue LED matrix |
| Sensors | None onboard |
| I/O Pins | UNO shield-compatible headers: 18-pin digital @3.3V (SPI, I2C, UART, PWM, CAN), 14-pin analog (ADC), Qwiic (I2C), 2x 60-pin high-speed bottom connectors |
| Dimensions | 68.85mm x 53.34mm (UNO form factor) |
| Use Case | Edge AI/ML, machine vision, robotics (Debian Linux + Zephyr RTOS) |
Full Tech specs: (Source)
NVIDIA Jetson Nano
| Device | NVIDIA Jetson Nano |
|---|---|
| Processor | 4-Core ARM Cortex A57 MPCore |
| Accelerator | 128-core Maxwell GPU (472 GFLOPS) |
| Memory | 4GB 64-bit LPDDR4 RAM |
| Storage | MicroSD card (16GB UHS-1 minimum) |
| Power Consumption | 20W (5V, 4A ≈ 3870 mW) |
| Connectivity | None specified (Ethernet, USB supported) |
| Camera/Display Support | 8MP (up to 3840 x 2160), Full HD 1080p+ |
| Sensors | None |
| I/O Pins | Not specified |
| Dimensions | Not specified |
| Use Case | AI applications, neural networks, image processing |
Full Tech specs: (Source)
For real-world data collection and testing.
(Lego sets) Robotics Platforms
Visuo XS809 Foldable Quadcopter
| Device | Visuo XS809HW |
|---|---|
| Camera | 2MP CMOS, 720p HD video |
| Camera FOV | 120° wide angle |
| Camera Mount | Manually tilt-adjustable, no gimbal (no mechanical stabilisation) |
| Video Transmission | WiFi FPV to smartphone app |
| Data Egress | App only — no CSI, USB or SD output |
| Gyroscope | 6-axis |
| Motors | 4x brushed (coreless) |
| Control Channels | 4 (Mode 2 controller) |
| RC Range | 80-100 m |
| Flight Time | 6-10 min per battery |
| Battery | 3.7V, 650-900mAh (30-60 min charge) |
| Dimensions | 31.5 x 31.5 x 13 cm unfolded; ~18 x 13 x 6 cm folded |
| Flight Modes | Headless, G-sensor, altitude hold, one-key return, 3-speed |
| Use Case | Aerial scene scouting, FPV situational awareness |
Full Tech specs: (Source)
Warning
The onboard camera is a closed FPV appliance — video is streamed over WiFi into a proprietary phone app with no CSI, USB or removable-media path off the aircraft, no frame-accurate timestamping and no exposure control. It cannot satisfy this repo's dataset provenance rules. For real aerial data collection, treat the airframe as a carrier and mount an independent payload with its own storage and clock. Full sensor analysis: Vision Sensors — Visuo XS809 FPV Camera.
Explore a range of Edge AI benchmarks, including both custom and industry-standard options like MLPerf, all tested on real hardware to ensure real-world performance.
Note
Coming soon