IAS100 AI Video Analytics Service Platform
The IAS100 AI Video Analytics Service Platform (hereinafter referred to as IAS100) is independently developed by Hope-Wish. The platform adopts intelligent analytics algorithms that combine traditional machine learning with deep learning, enabling efficient extraction of structured information from targets such as humans, vehicles, vessels, drones, and animals.
It provides reliable intelligent analysis results for application systems and delivers smart capabilities for 24/7 unattended monitoring scenarios, significantly reducing manpower and operational costs.
The platform is widely used in applications such as human and vehicle detection; fire and smoke detection in forests, oil depots, chemical plants, and industrial parks; low-altitude drone detection in airports, nuclear power plants, and urban areas; and vessel detection in ports, docks, lakes, and maritime environments.
Product Details
Key Features:
1.Enhanced Moving Target Detection: Supports advanced detection of both large and extremely small moving targets through target recognition and trajectory tracking; enables tripwire detection and region intrusion detection.
2.Diverse Human Action Detection: Supports detection of humans in various postures and movements, including crouching, walking, running, jumping, crawling, and swimming, for both large and small targets.
3.Multi-Class Target Detection: Supports detection of various targets, including humans, vehicles, vessels, animals, birds, drones, airborne objects, aircraft, and landing gear.
4.Dual-Spectrum Video Detection: Supports synchronized and asynchronous detection using visible light and thermal imaging, including same-rule and different-rule configurations.
5.Long-Range Small Target Detection: Supports intrusion alarm detection for targets smaller than 5×5 pixels; supports detection and recognition of ultra-small targets smaller than 15×15 pixels.
6.Cloud-Edge-Device Deployment: Supports integrated system solutions combining software and algorithms; supports deployment models including back-end analytics platforms, intelligent algorithm suites, and licensed algorithm models.
Application Scenarios:
Widely used in human and vehicle detection; fire and smoke detection in forests, oil depots, chemical plants, and industrial parks; low-altitude drone detection in airports, nuclear power plants, and urban areas; and vessel detection in ports, docks, lakes, and maritime environments.
Product Specifications
Model Name
IAS100 AI Video Analytics Service Platform
Applicable Devices
Electro-optical turntables, electro-optical PTZ units, electronic sentry devices, general security cameras
System Performance
- A single service supports no fewer than 16 channels of real-time video analysis (actual capacity depends on hardware configuration)
- A single service supports no fewer than 100 channels of time-shared video analysis (actual capacity depends on hardware configuration)
Image Analysis
1.Target Tracking: Supports single-target and multi-target tracking algorithms with real-time trajectory display
2.Panoramic Stitching: Supports visible light panoramic stitching and thermal imaging panoramic stitching
3.Semantic Segmentation: Supports semantic segmentation for both visible light and thermal imaging video
4.Moving Target Detection: Supports motion detection, tripwire detection, and region intrusion detection
Moving Target Detection Capability
- Supports detection of targets ≥ 5×5 pixels at 1080P resolution (visible light)
- Supports detection of targets ≥ 3×3 pixels at D1 resolution (thermal imaging)
- Detection accuracy after enhancement reaches over 95%
Ground Target Recognition Service (Optional)
1.Human Detection: Supports detection of human targets in visible light and thermal imaging video
2.Supports face detection in visible light and thermal imaging video
Detection Capability
- Visible Light (1080P):
Minimum detectable target ≤ 10×10 pixels; maximum ≥ 300×240 pixels
- Thermal Imaging (D1):
Minimum detectable target ≤ 8×8 pixels; maximum ≥ 200×160 pixels
- Detection accuracy ≥ 96% when human, vehicle, or animal targets exceed 30×30 pixels
- Detection accuracy ≥ 90% when fire/smoke targets exceed 50×50 pixels
- Single GPU supports ≥ 100 FPS processing speed
- A single service supports no fewer than 4 GPUs simultaneously
Water Surface Target Recognition Service (Optional)
1.Vessel Detection: Supports detection of vessels in visible light and thermal imaging video, including passenger ships, cargo ships, fishing boats, speedboats, naval vessels, and sand dredgers
2.Supports detection of swimming humans on water surfaces via thermal imaging
Detection Capability
- Visible Light (1080P):
Minimum detectable target ≤ 10×10 pixels; maximum ≥ 300×240 pixels
- Thermal Imaging (D1):
Minimum detectable target ≤ 15×15 pixels; maximum ≥ 200×160 pixels
- Detection accuracy ≥ 95% when vessel targets exceed 30×30 pixels
- Detection accuracy ≥ 85% for swimming and fishing targets above 35×35 pixels
- Detection accuracy ≥ 85% for floating objects above 50×50 pixels
- Single GPU supports ≥ 100 FPS processing speed
- A single service supports no fewer than 4 GPUs simultaneously
Low-Altitude Target Recognition Service (Optional)
1.Drone Detection: Supports detection of drones in visible light and thermal imaging video, including quadcopters, hexacopters, and fixed-wing UAVs
2.Airborne Object Detection: Supports detection of airborne objects such as kites and balloons (visible light), and balloons (thermal imaging)
Detection Capability
- Visible Light (1080P):
Minimum detectable target ≤ 10×10 pixels; maximum ≥ 300×240 pixels
- Thermal Imaging (D1):
Minimum detectable target ≤ 8×8 pixels; maximum ≥ 200×160 pixels
- Detection accuracy ≥ 95% when drone targets exceed 30×30 pixels
- Detection accuracy ≥ 80% when airborne object targets exceed 30×30 pixels
- Single GPU supports ≥ 100 FPS processing speed
- A single service supports no fewer than 4 GPUs simultaneously
Video Processing
1.Stream Decoding: Supports video decoding via direct device streaming and media server forwarding
2.Video Encoding: Supports combined encoding of decoded video and inference data
Hardware Adaptation
1.Supports algorithm migration and adaptation across various chips and GPUs, including HiSilicon, Atlas, Cambricon, and NVIDIA
2.Supports deployment across cloud, edge, and device environments
Interfaces & Protocols
1.Access Protocols: Supports RTSP and GB28181 (optional) video protocols; supports integration with mainstream manufacturers such as Hikvision, Dahua, and Uniview
2.Secondary Development: Supports SDK-based development; supports RESTful API integration; supports RTSP output of inference video streams
