←API Reference ←NSDK

SceneSegmentationResult

↳ inherits from AwarenessImageResult

---

Declaration

class SceneSegmentationResult

Summary

Contains semantic segmentation results from the NSDK scene segmentation processing system. SceneSegmentationResult provides semantic understanding of the environment by classifying pixels in camera images into different object categories (e.g., sky, ground, buildings, people, vehicles). This enables applications to understand the scene structure and make intelligent decisions based on environmental context.

Overview

Semantic segmentation results include:

• Confidence maps: Per-pixel confidence scores for semantic classifications
• Packed channels: Multiple semantic categories encoded in a single image
• Suppression masks: Masks indicating areas to be ignored or suppressed
• Metadata: Frame information, timestamps, and error status

Example Usage

// Get confidence for a specific semantic channel
let (status, confidenceResult) = sceneSegmentationSession.getLatestConfidence(channelIndex: 0)
if status.isOk(), let result = confidenceResult {
    print("Confidence image size: \(result.image?.width ?? 0) x \(result.image?.height ?? 0)")
    print("Frame ID: \(result.frameId)")
    print("Timestamp: \(result.timestampMs)")
    // Process confidence data for semantic understanding
    processSemanticConfidence(result)
}
// Get packed semantic channels
let (status, packedResult) = sceneSegmentationSession.getLatestPackedChannels()
if status.isOk(), let result = packedResult {
    // Process packed semantic data
    processPackedSceneSegmentation(result)
}

---