SenseCAP A1101 LoRaWAN Vision AI Sensor is a TinyML Edge AI-enabled smart image sensor. It supports a variety of AI models, such as image recognition, people counting, target detection, meter recognition, etc. It also supports training models with TensorFlow Lite.






Update


We're excited to have Edge lmpulse officially support our Grove-Vision Al Module. Now, you can follow our Wiki or go to the Edge lmpulse website to get started training and deploying your ML models!



 




Customize Vision AI Sensor with Seeed

If you would like to customize a Vision AI sensor with Seeed for different AI application scenarios, please feel free to submit a form here. It is a 5-minute survey for Seeed to better understand your customization needs. We will contact you to further discuss your needs if needed.



The SenseCAP A1101 - LoRaWAN Vision AI Sensor combines TinyML AI technology and LoRaWAN long-range transmission to create a low-power, high-performance AI device solution for both indoor and outdoor use. This sensor features Himax's high-performance, low-power AI vision solution, which supports training with Google TensorFlow Lite, Edge Impulse, and multiple TinyML AI platforms.

SenseCAP A1101 - LoRaWAN Vision AI Sensor is an image recognition AI sensor designed for developers. It will be your choice to open the door to the new TinyML world if you are:

- a machine learning developer who wants to develop edge AI and vision recognition applications and is finding a low-cost and reliable hardware platform to get started; or

- a software/hardware engineer who is keen to learn about TinyML, the model training process, and the algorism; or

- an enthusiast simply wants to experience the AI machine learning product without writing a line of code; or

- a SenseCAP user who wants to add a vision sensor to meet a specific application need; or

- a solution integrator or business developer who has specific AI application demands that need to deploy and scale in a vertical industry using a low-cost LoRaWAN network;

What can SenseCAP A1101 do?


SenseCAP A1101 is a product that needs developers' secondary development to deploy in a real-world application, and it enables users:


1) to test SenseCAP A1101 with the two demo models provided in the product (people detection and meter recognition) and experience Vision AI application;


2) to collect images using SenseCAP A1101 and generate a dataset through an annotate platform (e.g.: RoboFlow) ;


3) to train and generate their models through third-party AI training tools and deploy them to SenseCAP A1101;


4) to cooperate with an AI industry partner on dataset generation, model training, algorithm development, etc., to get customized and deployable models for specific application usages and deploy them in the SenseCAP A1101



AI Model Training Process


With operating temperature from -40 to 85℃ and IP66 waterproof grade, it is suitable for outdoor and harsh environments IoT applications. Meanwhile, we provide the SenseCAP all-in-one solution for this device which supports the rapid configuration of the Bluetooth APP, fast connection to gateways, and Cloud services.

Travel Data through SenseCAP Cloud for Free


With SenseCAP Mate App and SenseCAP Cloud, it is effortless to get and utilize data with few clicks, and they also support wide compatibility with other third-party tools. Through the plain and easy tutorials, you can connect the sensors through the SenseCAP Cloud and travel the data by LoRa® within 10 minutes.

TinyML Demo Models Provided for Quick Test


Embedded Human Body Detection and Meter Recognition TinyML model demos for a quick test and product experience.

IoT into the Wild


The theme of Seeed's new product launch in 2022 is "IoT into the Wild". Check low-power sensors to ultra-long-range gateways fully featuring Edge AI and LoRaWAN.


【About SenseCAP】Among the first launch of the Seeed industrial IoT (IIoT) product series, SenseCAP is focusing on wireless environmental sensing applications: smart agriculture, precision farming, and smart city, to name a few. It consists of hardware products (sensors, data loggers & gateways, etc.), software services (SenseCAP portal, mobile App, open dashboard), and an API for device & data management.






FAQs


What is the detailed specification of the camera in SenseCAP A1101?


HiMax HM0360 Sensor; Lens size 1/6"; 3P+1R Lens Construction; F/No 2.4±5%; Full-color 30W pixel; Max Frame rate 640*480 VGA 60 FPS (192*192 is the ideal supported resolution for AI model); Optical FOV 82º; EFL 1.98mm
If you have a need to customize the camera, please feel free to submit a form here.

What is the ideal camera setup for image collection?


For a 30W pixel camera, it is suggested to collect images within 1m-5m to get the ideal image effect. Please make sure to face the camera straight towards the objects and have sufficient illumination (no reflect light, no overexposure), and avoid large movement and vibration when collecting.

How far could the SenseCAP Vision AI camera detect an object and have good results?


Based on our test for the people detection model, we could reach results of 70% degree of confidence within 1m-5m.

Is it possible to change the camera or have a customized version SenseCAP Vision AI sensor?


Yes, and if you have large quantity demands, please contact us.


How long does the battery last?


It will highly depend on the image collection and data upload frequency. Based on our test, if the A1101 collects image and upload per minute, the battery could last for 9.5 months.


Is the battery rechargeable? Does it support solar energy charging?


No. If you have large quantity demands on the customized version, please contact us.


What could I do to improve and have a better result?


The AI result depends on many factors, including the quality of the images, the accuracy of the annotation, the way to generate the dataset, the model training process, the algorism used, etc.

The one you could improve with SenseCAP A1101 Sensor is to directly use SenseCAP A1101 camera to collect the image instead of using the existing images shot by other devices or using the existing dataset purchased from the third party. Especially when you are having a relatively small dataset, it will help a lot to improve the result.

And check your image collections to make sure they are having the right exposure (avoid insufficient or too strong light), right angle towards the objects, etc.

What is the difference between Grove Vision AI and SenseCAP A1101?



  • Grove Vision AI Module, which is provided in the SenseCAP K1100 - The Sensor Prototype Kit and needs to be developed with a mainboard with MCU (e.g.: Wio-Terminal or Arduino Uno). The Grove Wio-E5 module is required to connect to the mainboard if LoRaWAN transmission is needed; If the low power consumption feature is not the main consideration, it could also connect to the Cloud via WiFi. (Click here to learn more)

  • SenseCAP A1101 - LoRaWAN Vision AI Sensor is an industrial-level SenseCAP sensor that has an IP66 grade industry-level enclosure and battery, which has a Wio-E5 LoRaWAN module and MCU embedded. Benefiting from the reliable SenseCAP sensor platform, it is very convenient to add SenseCAP A1101 to the SenseCAP Mate App and Dashboard and deploy the application when the model training is completed.



Will there be more AI-related sensors developed by SenseCAP?


Now we only have SenseCAP A1101 Vision AI Sensor on the list. The next would be SenseCAP A1102 - LoRaWAN Sound/Vibration AI Sensor, but at the moment it is still under development. Please stay tuned.




Part List


SenseCAP A1101 - LoRaWAN Vision AI Sensor *1