Edge AI Box with 20+ AI algorithms and 16 channels of 1080P video input. Each channel can run 2 AI algorithms simultaneously, allowing a total of 18 algorithms to operate concurrently.
■ High performance embedded Linux operation system.
■ 20+ diverse and enriched AI algorithms.
■ Works with any camera RTSP,ONVIF,GB/T28181 stream to add real-time deep learning analytics to your existing camera system.
■ Perfectly integrated in Daxun Digital video surveillance management software SVMS Pro.
■ Support 16ch 1080P video input, each channel can be configured with up to 2 algorithm functions, totally up to 18 algorithm functions can be operated concurrently.
Available model: AIBOX-H16
Daxun Digital AI Box H Series is a economic edge AI computing series products, that empowers your system with unparalleled AI video analytics capabilities. Seamlessly integrating with cameras, it provides access to real-time video streams of normal IP cameras, triggers alarms and analyzes events data after intelligent analysis rules have been configured.Thus it offers a standalone solution for comprehensive video analysis and supports detection of different types of events, such as tripwire, intrusion, smoking, fight, and crowd gathering etc. Additionally, the deep integration with Daxun Digital video management software SVMS Pro is offering for a truly unified experience,while it also effortlessly transfers analyzed results to third-party platform through network (IAN and WAN).
■ High performance embedded Linux operation system.
■ 20+ diverse and enriched AI algorithms.
■ Works with any camera RTSP,ONVIF,GB/T28181 stream to add real-time deep learning analytics to your existing camera system.
■ Perfectly integrated in Daxun Digital video surveillance management software SVMS Pro.
■ Support 16ch 1080P video input, each channel can be configured with up to 2 algorithm functions, totally up to 18 algorithm functions can be operated concurrently.
| Model | AIBOX-H16 | ||
|---|---|---|---|
| Camera Access (Provide access to and perform analysis on the camera.) |
A single AI box supports up to 16 channels of 1080p cameras at the same time. | ||
| System | Linux | ||
| CPU | 8-Core Arm CA53@1.6GHz | ||
| Memory | 8G | ||
| EMMC | 32G | ||
| Video Decoding | H.264&H.265:1080P@480fps 8192*8192/8K/4K/1080P/720P/D1/CIF | ||
| Video Encoding | H.264&H.265:1080P@240fps8K/4K/1080P/720P/D1/CIF | ||
| Image Encoding/Decoding | JPEG:1080p@240fps, maximum 32768*32768 | ||
| Interface | 2*RJ 45 10M/100M auto adaptive; 2*USB3.0/HDMI2.0/Audio(In/Out)/CAN/RS-232/RS-485/GPIO/SATA3.0 | ||
| Interface(Optional) | LTE/5G/WI-FI/BT | ||
| Power Consumption | 15W | ||
| Power Supply | AC 220V | ||
| Dimensions (mm) | 219.8mm*200mm*46.8mm | ||
| Working Temperature | -20℃~+60℃ | ||
| Protection Level | IP41 | ||
| Algorithm Functions | |||
| Flame and Smoke Detection | When detect the target smokes and flame in the image, an alarm will be triggered. (Target pixel≥ 150 x 150, recommended light intensity: 70lux~900lux .) | ||
| Safety Helmet Detection | Detect someone is not wearing safety helmet in the video, an image will be captured and an alarm will be triggered. This detection is usually used on construction sites and production workshops to prevent safety accidents due to not wearing safety helmets. (Target pixel≥ 150 x 300.) | ||
| Reflective Vest Detection | Detect whether the person in the image is wearing the reflective clothes. If the person is not wearing the reflective clothes, an alarm will be triggered and the image will be captured as well. This detection is usually used on the construction site to detect whether the construction workers wear reflective clothes, so as to ensure the safety during the construction process.(Target pixel≥ 150 x 300.) | ||
| Bare Arms Detection | Detect whether the person in the image is wearing long-sleeved clothing. If the person is wearing short-sleeved T-Shirt and leave his arms exposed, an alarm will be triggered and the image will be captured as well. (Target pixel≥ 150 x 300.) | ||
| Mask Detection | Detect whether person in the image is wearing the mask. If the person is not wearing the masks, an alarm will be triggered and the image will be captured as well. This detection is used for the places which need to wear the masks such as the kitchen room of the restaurants, schools’ kitchen room etc. (Target pixel≥ 150x300.) | ||
| Face Capture | Snapshot the face image in the video, up to 20 faces can be detected simultaneously. (Target pixel≥ 150x150.) | ||
| Away-From-Post Detection | Set monitoring zones to detect when people on the position is less than the defined number of people and exceed defined duration of time. You can configure the defined number of people and defined duration of time not to exceed, the polygons to mark out detection zones. (Target pixel≥ 150x150.) | ||
| Line Crossing Detection | Draw lines to establish tripwire zone for detection. Once the line is crossed, it will trigger an alarm, and the image will be captured as well. (Target pixel≥ 150x300.) | ||
| Area Intrusion Detection | Set monitoring zones to detect when a person has entered the monitored zone. Once the zone is intruded, it will trigger an alarm, and the image will be captured as well. You can use polygons to mark out detection zones. (Target pixel≥ 150x300.) | ||
| Touch Fence | Set monitoring zone to detect people crossing the fence. When detect, an alarm will be triggered and the image will be captured as well. | ||
| People Counting | Draw line to establish counting zone for detection. When people cross the line, the number will be counted. If no one in 10S will cross the line, the counting number will reset to 0. (It supports minimum pixel 60*60, and recommended target pixel≥ 150x150.) | ||
| Smoking and Phone Call Detection | Detect people is smoking or playing the phone in image. An alarm will be triggered and an image will be captured. Phone and cigarette butts should be visible and appear on the upper body. (It supports minimum pixel 100*200, and recommended target pixel≥ 150x300.) | ||
| Face Recognition | Identify the face in the image, extract facial features and compare them with face picture in the database. When the similarity exceeds the set value, it will snapshot and report the facial inform. When the similarity is below the set value, report it as a stranger. The face database supports 20000 face pictures. (It support minimum pixel 120*120, and recommended target pixel≥ 150x150). It needs camera to be able to clearly capture the front side of person’s face. Picture for face database requirements: the pitch angle, yaw angle, and roll angle do not exceed 45°, without obstruction. | ||
| Vehicle no-parking detection | Draw polygons to mark out no-parking detection zones. If the cars enter this zone, the alarm will be trigger and the picture will be captured. If the car leaves the zone , then come back again. The new alarm will be trigger as well. (It supports minimum pixel 100*120, and recommended target pixel≥ 150x300). | ||
| Little Animal Detection | Detect little animal such as dog, cat, rat etc in the image, an alarm will be triggered, and a picture will be captured as well. | ||
| Wander Detection | Set monitoring zones to detect when a person stays or loiters in a monitored zone for longer than the defined period, an alarm will be triggered and a picture will be captured. You can use polygons to mark out detection zones, and set defined time. | ||
| Climbing Detection | Detect the person in the image doing climbing behavior, an alarm will be triggered and a picture will be captured. (It supports minimum pixel 100*200, and recommended target pixel≥ 150x300). | ||
| Fight Detection | Detect when a fight has broken out, an alarm will be triggered and the image will be captured. (It supports minimum pixel 100*200, and recommended target pixel≥ 150x300). | ||
| Uniform Detection | Detect whether the person in the image is wearing working uniform. If not wearing working uniform, an alarm will be triggered, a picture will be captured. (It supports minimum pixel 120*120, and recommended target pixel≥ 150x150). | ||
| Sleeping-Post Detection | Detect person sleeping on the table, an alarm will be triggered, a picture will be captured. ( It supports minimum pixel 100*120, and recommended target pixel≥ 150x150). | ||
| People Fall Down Detection | Detect people falling down in the image, an alarm will be triggered and an image will be captured. | ||
| Passageway Occupancy Detection | Detect whether there are road-occupying operations in the image, such as awnings, parasols etc. (It support minimum pixel 120*120, and recommended target pixel≥ 150x150). | ||
| Electric bicycle detection | Detect whether there is an electric bicycle in the elevator. When detect it, an alarm will be triggered. | ||
AI box datasheet downloads: click here to download.
Products
Contact Us Email: daxunview@gmail.com
Phone: +8618857869522
Tel：+86-0571-88088003
Teams:+8618857869522
Wechat:+8618857869522
WhatsApp:+8618857869522
Address: Room 316,Zi Jin Gang Commercial Building,Gu Dun Road No.656, Hangzhou, Zhejiang Province, China
