AI Object Measurement from Photos in Mobile Apps

A client wants to measure the length of a sofa from a single photo—without a tape measure or special equipment. We implemented this feature in mobile apps with accuracy up to 2% under proper conditions. Our experience: 5+ years in iOS/Android development, 30+ projects with AR and computer vision. We offer a turnkey solution: from method selection (LiDAR, SLAM, or monocular depth) to integration with your backend. Get a consultation—we'll evaluate your project in one day.

Which Measurement Method to Choose: LiDAR, SLAM, or Monocular Depth?

ARKit/ARCore (LiDAR or SLAM) is accurate but requires device support. iPhone 12 Pro and newer with LiDAR deliver 1–3 cm accuracy at distances up to 5 meters. ARCore on Android without LiDAR is worse, with 3–8 cm error.

Monocular depth estimation works on any device without LiDAR, using CNN to estimate depth from a single frame. MiDaS, DPT, Depth Anything V2 are current models. Accuracy is noticeably lower than LiDAR but sufficient for many tasks.

// iOS: method selection based on device capabilities
func selectMeasurementMethod() -> MeasurementMethod {
    if ARWorldTrackingConfiguration.supportsSceneReconstruction(.mesh) {
        return .lidarARKit          // iPhone 12 Pro+, iPad Pro
    } else if ARWorldTrackingConfiguration.isSupported {
        return .slamARKit           // ARKit without LiDAR
    } else {
        return .monocularDepth      // fallback to CoreML model
    }
}

What Affects Measurement Accuracy?

Measurement accuracy directly depends on shooting conditions: lighting, surface texture, distance to the object, and camera angle. White walls without texture degrade SLAM tracking, and high lighting can cause overexposure. For stable results:

• Textured surfaces (black, glossy objects are harder to detect)
• Distance to the object no more than 5 m with LiDAR, 3 m with SLAM
• Minimize dynamic objects in the frame
• Proper camera focus

How We Implement Measurement via ARKit

// Measuring distance between two points in AR
class ARMeasurementSession: NSObject, ARSessionDelegate {

    var arView: ARSCNView!
    private var startAnchor: ARAnchor?
    private var endAnchor: ARAnchor?

    func placePoint(at screenPoint: CGPoint) -> MeasurementPoint? {
        // Raycast from screen to 3D world space
        guard let query = arView.raycastQuery(
            from: screenPoint,
            allowing: .estimatedPlane,
            alignment: .any
        ) else { return nil }

        guard let result = arView.session.raycast(query).first else { return nil }

        let worldPosition = result.worldTransform.columns.3  // position in meters
        return MeasurementPoint(
            position: SIMD3(worldPosition.x, worldPosition.y, worldPosition.z),
            confidence: result.targetAlignment == .horizontal ? .high : .medium
        )
    }

    func calculateDistance(from start: MeasurementPoint, to end: MeasurementPoint) -> Measurement<UnitLength> {
        let diff = end.position - start.position
        let distanceMeters = Double(simd_length(diff))
        return Measurement(value: distanceMeters, unit: .meters)
    }
}

A common mistake is not accounting that raycast works best on well-textured surfaces. A white wall produces poor SLAM tracking, causing AR markers to drift.

Displaying the Measurement in AR

func addMeasurementLine(from start: SIMD3<Float>, to end: SIMD3<Float>,
                         distance: String) {
    let midpoint = (start + end) / 2

    // Line between points
    let lineNode = SCNNode(geometry: createCylinder(from: start, to: end))

    // Label with distance at midpoint
    let labelNode = SCNNode(geometry: SCNText(string: distance, extrusionDepth: 0.001))
    labelNode.position = SCNVector3(midpoint.x, midpoint.y + 0.02, midpoint.z)
    labelNode.scale = SCNVector3(0.005, 0.005, 0.005)
    labelNode.constraints = [SCNBillboardConstraint()]  // always face the camera

    sceneRoot.addChildNode(lineNode)
    sceneRoot.addChildNode(labelNode)
}

Reference Object Approach for Photo Measurement

Without AR—an object of known size in the frame is needed. A bank card (85.6 × 53.98 mm) is a convenient reference:

// Android: measurement via reference object
class ReferenceObjectMeasurer {

    fun measureWithCard(bitmap: Bitmap, cardBoundingBox: RectF,
                        objectBoundingBox: RectF): MeasurementResult {
        // Real card dimensions
        val cardRealWidth = 85.6f  // mm
        val cardRealHeight = 53.98f

        // Pixels → mm
        val pixelsPerMmHorizontal = cardBoundingBox.width() / cardRealWidth
        val pixelsPerMmVertical = cardBoundingBox.height() / cardRealHeight

        // Perspective distortion correction (simplified)
        val correctionFactor = estimatePerspectiveCorrection(
            cardBoundingBox, imageDimensions = bitmap.width to bitmap.height
        )

        return MeasurementResult(
            widthMm = (objectBoundingBox.width() / pixelsPerMmHorizontal) * correctionFactor,
            heightMm = (objectBoundingBox.height() / pixelsPerMmVertical) * correctionFactor,
            accuracy = MeasurementAccuracy.MODERATE  // ±5-10% without calibration
        )
    }
}

Card detection in the frame is done via ML Kit Object Detection or a custom YOLOv8 model (easy to train on 500 card photos in various conditions).

Process: From Idea to Release

1. Analytics — study business requirements, audience devices, required accuracy.
2. Design — choose architecture: native ARKit/ARCore or cross-platform, define measurement method (LiDAR/SLAM/monocular depth).
3. Implementation — develop MVP with basic UI (two points, line), integrate CoreML/ML Kit, configure calibration.
4. Testing — verify accuracy on 10+ devices under different shooting conditions, A/B tests.
5. Deployment — prepare for App Store and Google Play release, documentation, TestFlight/Firebase Distribution.

What's Included in the Work

• Architectural documentation and API description
• Integration with your CRM/backend via REST or GraphQL
• Training your team on using the feature
• 3 months of technical support
• Access to source code (with modification rights)
• Certificates from Apple and Google (if required)

We are certified Apple and Google developers. We guarantee measurement accuracy in compliance with App Store Review Guidelines (Section 4.2/5.1). Over 5 years, we have delivered 30+ projects in e-commerce, construction, and healthcare.

Evaluate your project—write to us. Get a consultation on method selection and implementation timeline.

This article references monocular depth estimation and Apple ARKit documentation.

Machine Learning in Mobile Apps: CoreML, TFLite, and On-Device Models

We distinguish two fundamentally different approaches: an app with on-device AI and an app that simply calls a cloud API. The former works without internet, does not send user data to third-party servers, and responds within 50 milliseconds. The latter depends on network latency and pricing plans. Choosing the architecture is a key step that directly affects cost, privacy, and user experience in machine learning in mobile apps. Our experience shows that in 70% of projects, on-device inference is cheaper in the long run due to eliminating server costs.

How to Choose Between CoreML and TFLite for On-Device Inference?

CoreML — Apple's native framework for running ML models on device. Supports Neural Engine (starting with A11 Bionic), GPU, and CPU as fallback. Models are converted to .mlmodel format via coremltools from PyTorch, ONNX, or TensorFlow. Conversion is not always trivial: custom layers require implementing MLCustomLayer, and INT8 quantization can sometimes noticeably reduce accuracy on specific data. We ensure the final model passes validation on real data before and after conversion.

TensorFlow Lite — cross-platform alternative for Android and Flutter. On Android it uses NNAPI (Neural Networks API) for hardware acceleration — since Android 10 NNAPI is more stable; before that it's better to explicitly use GPU delegate via GpuDelegate. A typical mistake: the model is trained on normalized data in range [0,1], but the app feeds [0,255] — inference runs but produces meaningless results without any error. We include an automatic input data validation module in the SDK.

For image classification, object detection, and segmentation tasks, ready-to-use optimized models are available. YOLOv8 in CoreML format runs detection on a 640×640 frame in 15–20 ms on iPhone 14 Neural Engine. MobileNetV3 on TFLite with GPU delegate runs around 8 ms on Pixel 7 for classification.

What If You Need Text Generation On-Device?

Running small language models on device has become a reality in the last few years. Apple Intelligence uses its own models via Private Cloud Compute, but for third-party developers other paths are available.

llama.cpp with Metal backend on iOS is a working approach for phi-3-mini (3.8B parameters, 4-bit quantization, ~2.3 GB). Inference: 15–25 tokens/second on iPhone 15 Pro. For integration in Swift, use the Swift Package llama.swift or a wrapper via C interface llama.h. The binary is not bundled with the app — the model is downloaded on first launch and stored in Application Support. Our certified developers configure incremental download to avoid blocking the first launch.

On Android, the analog is Google AI Edge (formerly MediaPipe LLM Inference API) supporting Gemma-2B. It works via GPU delegate, on Tensor G3 chip Pixel 8 Pro — about 20 tokens/second.

Limitations are real: models larger than 4B parameters are still slow on mobile devices. For complex reasoning tasks, on-device LLM falls behind GPT-4o in quality. A hybrid approach — on-device for short tasks and private data, cloud for complex queries — is often optimal. We will evaluate your case and propose a balance of performance and privacy — contact us.

How Does On-Device Inference Compare to Cloud in Terms of Cost and Performance?

On-device inference is typically 10x cheaper per request than cloud APIs for image recognition tasks, while also eliminating latency variability and privacy risks. The table below summarizes the trade-offs.

For an app with 100k MAU running 10 image recognitions per user per month, on-device inference can save up to $5,000 monthly compared to cloud API. Get a free consultation on your ML architecture today.

Integrating OpenAI API and Other Cloud Models

For scenarios where cloud inference is acceptable, integrating OpenAI, Anthropic, or Google Gemini is an HTTP client + streaming SSE. In Swift, AsyncThrowingStream is convenient for streaming responses. In Kotlin, use Flow.

Critically: API keys must never be stored in the app bundle. Even an obfuscated key can be extracted from the IPA in 10 minutes using strings or frida. Correct architecture: mobile app → your own backend → OpenAI API. The backend controls rate limiting, logs requests, and protects the key.

What Is Included in the Work (Deliverables)

• Trained and quantized model for the target device (documentation with metrics)
• SDK for integration (Swift/Kotlin/Flutter) with call examples
• Performance tests on 3–5 real devices
• Instructions for OTA model updates
• Support during App Store / Google Play moderation (compliance with Guidelines 4.2, 5.1)
• 2 weeks of technical support after release

Typical Project Pipeline

1. Task analysis — measure latency, privacy, size, supported devices.
2. Model prototyping — in Python, evaluate accuracy on target data.
3. Conversion and quantization — for CoreML/TFLite with validation.
4. Integration into the app — model wrapped in a service layer (easy to swap CoreML ↔ TFLite ↔ cloud).
5. Testing — on real devices, measure FPS, RAM, battery.
6. Deployment — via TestFlight / Firebase App Distribution, monitor metrics.

Timelines: integration of a ready CoreML/TFLite model — 1–2 weeks, development of a custom model with mobile optimization — from 6 weeks, on-device LLM chat with personalization — 4–8 weeks.

Why We Take on Complex Cases?

10+ years of experience in mobile development, 50+ implemented AI/ML solutions, guarantee of compatibility with current iOS and Android versions. All projects undergo code review and load testing. The cost includes preparation of moderation documentation and training of your team.

Contact us — we will help you choose the architecture and implement ML in your app turnkey. Order an audit of your existing solution — we will assess the potential for server cost savings free of charge. In some projects, savings can reach significant amounts per month.