BODY FEATURE EXTRACTION SDK
Detect people in images or decoded video frames and extract feature vectors from the detected person regions on your RK3588 device.
The SDK detects people, crops their image regions and returns numerical feature vectors. These describe image features, not a person’s name or an attribute report. Your application handles video decoding, result storage and downstream processing.
The delivered package defines the person class ID, input format, preprocessing, feature vector dimension and output data type. Use the specifications for that model version rather than assuming different models share the same output format.
Vehicle and bicycle detection, other additional detection classes, person attribute analysis and cross-camera person search are outside the standard Body SDK. Availability, supported classes and delivery scope must be verified separately before any additional module or project is agreed.
Evaluate Body SDK
Workflow illustrationEXPLORE THE DEMO
Open the 88PL experience platform. When you are ready to discuss integration, return here to request an SDK evaluation.
Open demo platform · Opens in a new tabOnline demos and an SDK evaluation are separate steps. Opens in a new tab
Integration details
GOOD TO KNOW
No. The standard SDK returns person detections and feature vectors. Cross-camera search also needs matching, a searchable data store and application logic, with its own validation. Those functions require a separately assessed project.
Vehicle detection is not included in the standard Body SDK scope. If you need it, specify the required object classes before evaluation. We will assess the model support and proposed delivery scope separately.
A feature vector is a numerical representation of an image region; it does not itself return a name or verify identity. Identity-related use requires a separate application and validation for the intended use case.
Start with your hardware, your requirements and a focused evaluation.