SenseCAP Vision AI Sensor A1102

Features

The SenseCAP A1102 LoRaWAN® Vision AI Sensor is a smart edge device that runs TinyML AI models locally, transmitting only inference results via LoRaWAN® (863-928MHz) for enhanced privacy and power efficiency. Operating at 480×480 resolution with 10 FPS, it comes with three pre-deployed models for human detection, people counting, and meter reading, while offering no-code access to 300+ pre-trained models or custom model training through the SenseCraft AI platform. It also supports model deployment from TensorFlow Lite and PyTorch frameworks. Users can configure the device and the frequency bands via the SenseCraft App, and deploy the IP66-rated sensor in harsh indoor or outdoor environments. With a built-in battery, making it ideal for various AI applications requiring long-range communication.

• Advanced Edge AI Camera: Equipped with a powerful MCU-based vision camera processor performing local inference at 480×480 resolution at 10 FPS. By transmitting only inference results, it ensures data privacy and minimizes network bandwidth usage.
• No-Code AI Model Deployment and Training: Pre-loaded with three on-device AI models (human detection, people counting, and meter reading) for out-of-box vision applications. Natively integrates with SenseCraft AI , a web-based platform offering 300+ pre-trained models for instant deployment and simple 3-step training workflow for classification and detection models. This enables rapid model training and deployment without coding expertise. Moreover, the devices also supports custom AI models converted from TensorFlow and PyTorch frameworks for edge deployment.
• LoRaWAN® Connectivity: Supports global frequency bands (863-928 MHz) with transmission range up to 10 km in optimal conditions, enabling reliable long-distance communication.
• 2 Power Options: Features a 19Ah SOCl2 D-size battery inside. Compatible with external DC power supply via optional junction box.
• IP66 Rating Environmental Durability: Features IP66-rated protection against dust ingress and strong water jets. Operates reliably in temperatures from 0°C to 70°C, suitable for diverse indoor and outdoor deployments.
• SenseCraft Cloud Integration: Includes SenseCraft Mate mobile app and web portal for easy device configuration, device management and quick data visualization. Features open API support and 3-month FREE trial access, it’s compatible with third-party IoT platforms.
  
  

Specifications

Power Supply Options

• Option 1: The SenseCAP A1102 operates in cyclic mode using a built-in 19Ah D-type Li-SOCl2 battery. The battery is commercially available for easy replacement and enables standalone outdoor deployment without requiring external power.
• Option 2: The SenseCAP A1102 supports constant operation through a 12V DC power supply connection. An extra matching junction box is needed for straightforward installation and assembly.
  
  

Hardware Overview

• Onboard
  
  
  

Applications

• Person Detection
• People Counter
• Face Detection
• Specific Object Detection

Part List

1. What are the ideal camera settings for image capture?

For a 5-megapixel camera, it is recommended to capture images within a range of 1m to 5m to achieve optimal image quality. During image capture, ensure that the camera is directly facing the subject, with sufficient lighting (no glare or overexposure), and avoid significant motion or vibration.

2. How far can the SenseCAP Vision AI camera detect objects and achieve good results?

Based on our testing with a human detection model, the camera can achieve a 70% confidence level for results within a range of 1m to 5m.

3.cc.

4. Can the battery be recharged? Does it support solar charging?

No.cc.

5. What should I do to improve and achieve better results?

AI results depend on many factors, including image quality, annotation accuracy, the way datasets are generated, model training processes, and the algorithms used.

One improvement for the SenseCAP A1102 sensor is to directly use the SenseCAP A1102 camera to collect images instead of using existing images captured by other devices or pre-purchased datasets from third parties. This approach, especially when you have a relatively small dataset, can significantly improve results.

Additionally, check your image collection to ensure proper exposure (avoiding underexposure or overexposure) and that the images are captured at the correct angles relative to the objects.

6. What are the differences between SenseCAP A1102 and SenseCAP A1101?

The SenseCAP A1102 has stronger performance compared to the SenseCAP A1101, offering better Vision AI capabilities, more supported models and frameworks, and the ability to transmit keyframe images via Wi-Fi.