Products
Your current location: Home > Products > AI computing power system
  • Introduction
  • Type
  • Features
  • Parameter
  • Case
Hengtong Edge Server is a high-density server designed for AI inference. Efficiently adapting to all AI algorithms in the market, realizing applications such as video structuring, facial recognition, behavior analysis, and status monitoring, empowering AI in fields such as smart cities, smart transportation, smart energy, smart finance, smart telecommunications, and smart industry.
HT-AI-TEES
● Support parking detection
● Support retrograde detection
● Support illegal lane change detection
● Supports simultaneous detection of 4 screens
● Support video file detection
● Supports MP4 and AVI recording formats
● Supports multiple data storage methods
● Support default adaptation province settings under low matching degree
● Support inclined travel detection filtering settings
● Support detection algorithm selection: full detection or lane change detection
● Supports ONVIF and RTSP video signal sources
● Support real-time preview of video images
● Support video group preview
● Support manual start and stop detection
● Support displaying illegal snapshot lists
● Support displaying illegally captured images
● Support video playback
● Support exporting images and video files
● Events can be queried based on time
● Events can be queried based on the video host
● Events can be queried based on the license plate number
● Events can be queried based on illegal status (parking, reversing, lane changing)
● Support system log query
● Support third-party reporting, can push pictures or videos
No. Project Parameters & performance
1 Operating system Embedded linux operation system
2 Master Chip Built in TPU AI chip with a computing power of 2TOPS
3 Algorithm type Deep learning algorithms
4 Operation interface WEB
5 Storage capacity 500G SSD
6 Chassis High performance heat dissipation aluminum material, fanless design
7 Compatibility Support ONVIF and RTSP video signal source access, compatible with mainstream domestic snapshot protocols
8 Video Decoding Capability 1 channel of high-definition video with a maximum resolution of 9 million pixels (4096 * 2160)
9 Single image operation inference time <36ms, real-time image analysis at 25 frames per second
10 Target tracking capability The maximum number of tracking targets for a single channel vehicle is greater than 150, and the tracking targets support a minimum pixel size of 64 * 64
11 Detection distance Detection distance ≥ 150 meters, detection width greater than 5 lanes, achieving no blind spot detection in the area
12 Event detection type Support capturing and collecting evidence of lane changes, parking, reverse, reversing, vehicle speed, following distance, congestion, and other situations. Support for extended spill detection and driver facial recognition
13 Storage function Support the capture of illegal scene images, snapshot line scene images, designated images of illegal vehicles, and images of illegal vehicle license plates (pre lane change images, post lane change images, close-up images, license plate images), linkage event recording, and support for offline continuation of transmission
14 Offline computing capability Support post analysis of illegal activities through historical video data, and support detection of multiple video formats such as MP4 AVI
15 Alarm input/output Input: 2 channels, triggered by high/low level signals;
Output: 4 channels, switch output
16 Network interface 2 RJ45 interfaces, dual gigabit network supports cross segment access
17 Power supply DC 12V
18 Power consumption Maximum<24W
19 Operation temperature -25℃~＋70℃
20 Weight 1kg
21 Working humidity 10％~95％
22 Chassis size 212mm(W)×200mm(D)×53mm(H)