A model is still running

A model is still running or downloading. Leaving will stop the current operation. Stay on this page to let it finish.

Skip to content
LocalAI.toolsBeta

Join the waitlist

Get an email when the new apps and model runs launch.

No account needed.

MVANet Desktop

Native MVANet still-image foreground cutout using pinned ONNX Runtime CPU or profiled CoreML GPU.

Desktop inferencePrivate · Runs on your computer
About this model

Native MVANet still-image foreground cutout using pinned ONNX Runtime CPU or profiled CoreML GPU.

License
MIT
Model released
Unknown
Source repository created
Mar 8, 2024
Model weights · FP32
422.0 MB

Best for

  • Portrait and object foreground cutout experiments
  • Background replacement

Keep in mind

  • Fine hair, transparency, unfamiliar subjects and poor input quality may produce incorrect foreground masks; not safety-critical segmentation.
  • Scoped portrait/dog quality only; no broad segmentation certification.
  • Atomic segmentation has no incremental output.
  • Verified full backup and cold-source recovery acceptance pending.
  • CoreML cold compilation took157–177 seconds in retained managed/helper tests; CPU initialization is substantially faster. Repeated warm GPU inference is faster than CPU.
  • Single forward pass converted graph; no equivalence claim to upstream flip/scale ensemble DIS5K evaluation.

Processed on your deviceYour input is processed locally in the desktop app. Model and runtime files download from external hosts.

Target platforms · verification required

  • macos · aarch64, x86_64
  • windows · x86_64

Recommended configuration

FP32 · cpu (verified platform only)

Sources & technical details
Original model
https://github.com/qianyu-dlut/MVANet
Release verification
Unknown
Release evidence
CVPR2024 original research; exact original checkpoint release day not established in bounded official review. Conversion date is separate.
Source updated
Feb 28, 2025
Native model files
onnx-community/MVANet-ONNX
Files updated
Apr 14, 2026
Artifact revision
43ec3427514b8d9164eea02df93ca2f1b036bb7b

Desktop app

Sample result

MVANet Desktop

Run MVANet Desktop on your computer

Download the desktop app to load this model, provide input and view results locally.

Public desktop downloads are coming soon.

Device requirements

macOS Apple SiliconCPU / FP32Tested16.0 GiBRAM estimate
macOS Apple SiliconCOREML / FP32Tested16.0 GiBRAM estimate
If your device falls short

MemoryAdvisory only: close other apps, reduce image/tile/token/batch settings or choose a smaller verified model/precision. CPU can need more RAM than GPU.

GPUIf the selected GPU runtime fails, existing CPU fallback may recover when enabled, but CPU has its own memory/runtime needs.

PlatformUse a compatible OS/architecture/runtime or another model. These estimates do not add hardware detection or block runs.

Benchmark evidence

Saved desktop measurements apply to the recorded hardware and runtime. Fallback cases report the device that actually ran; they do not prove GPU support on the requested device.

MVANet Desktop

12 configurations tested
Passed12
Failed0
TestedNov 16, 2026
Runtimeonnxruntime 1.23.2

Quality criteria

{ "frozenAt": "2026-11-16T13:20:00Z", "scope": "Existing independently authored foreground/background point annotations on two natural photographs. Portrait alpha IoU/MAE/boundaryF1 separately measures MODNet reference agreement, not human alpha truth.", "source": "research/slimsam-family-expansion/quality-gates.json", "sourceSha256": "2a1a8eddb1ab2cb5d82d99cc722f0d19d9375a90536b8a0a56fb9c36826abd4d", "minimumProbeAccuracy": 0.9, "maximumInferenceMs": 10000, "cases": [ { "file": "model-samples/modnet-portrait-pexels-5965592.jpg", "expected": { "width": 1024, "height": 683, "point": [ 470, 330 ], "minimumForegroundRatio": 0.15, "maximumForegroundRatio": 0.4, "foregroundPoints": [ [ 450, 120 ], [ 460, 210 ], [ 430, 350 ], [ 350, 420 ], [ 460, 520 ], [ 480, 630 ], [ 580, 470 ] ], "backgroundPoints": [ [ 50, 350 ], [ 800, 350 ], [ 900, 100 ], [ 100, 600 ], [ 700, 600 ], [ 200, 200 ] ], "minimumProbeAccuracy": 0.9 }, "source": "WebAI independently annotated whole-person foreground/background probes on the Pexels portrait, visually reviewed at original 1024x683 resolution.", "license": "Pexels License", "id": "person-box", "userInput": { "input_boxes": [ [ 230, 40, 740, 683 ] ] } }, { "file": "model-samples/desktop/mobilenet-dog.jpg", "expected": { "minimumForegroundRatio": 0.25, "maximumForegroundRatio": 0.8, "foregroundPoints": [ [ 200, 150 ], [ 350, 200 ], [ 500, 400 ] ], "backgroundPoints": [ [ 0, 0 ], [ 690, 570 ], [ 680, 10 ] ], "minimumProbeAccuracy": 0.9 }, "source": "https://commons.wikimedia.org/wiki/File:YellowLabradorLooking_new.jpg — Elf; derivative Djmirko/FT2; independent visual foreground/background probes.", "license": "CC-BY-SA-3.0", "id": "dog-box", "userInput": { "input_boxes": [ [ 50, 30, 655, 560 ] ] } } ], "familyId": "mvanet", "runtime": "Transformers.js4.3.2 preferred browser; pinned native Transformers architecture evaluated before ONNX/PyTorch alternatives", "preprocess": "Pinned processor bilinear1024/ImageNet normalization; exact graph signatures determine sigmoid handling. No BEN2 timing reuse." }

PrecisionDeviceResultPlatformInference / load
FP32coremlPassedFastmacos · aarch641339 ms / 156652 ms
FP32coremlPassedFastmacos · aarch641426 ms / 156652 ms
FP32coremlPassedFastmacos · aarch641444 ms / 156652 ms
FP32coremlPassedFastmacos · aarch641568 ms / 156652 ms
FP32coremlPassedFastmacos · aarch642186 ms / 156652 ms
FP32coremlPassedFastmacos · aarch644485 ms / 156652 ms
FP32cpuPassedModeratemacos · aarch645385 ms / 344 ms
FP32cpuPassedModeratemacos · aarch645424 ms / 344 ms
FP32cpuPassedModeratemacos · aarch645446 ms / 344 ms
FP32cpuPassedModeratemacos · aarch645660 ms / 344 ms
FP32cpuPassedModeratemacos · aarch645662 ms / 344 ms
FP32cpuPassedModeratemacos · aarch645822 ms / 344 ms
Test setup & sample details
Quality gate
{ "frozenAt": "2026-11-16T13:20:00Z", "scope": "Existing independently authored foreground/background point annotations on two natural photographs. Portrait alpha IoU/MAE/boundaryF1 separately measures MODNet reference agreement, not human alpha truth.", "source": "research/slimsam-family-expansion/quality-gates.json", "sourceSha256": "2a1a8eddb1ab2cb5d82d99cc722f0d19d9375a90536b8a0a56fb9c36826abd4d", "minimumProbeAccuracy": 0.9, "maximumInferenceMs": 10000, "cases": [ { "file": "model-samples/modnet-portrait-pexels-5965592.jpg", "expected": { "width": 1024, "height": 683, "point": [ 470, 330 ], "minimumForegroundRatio": 0.15, "maximumForegroundRatio": 0.4, "foregroundPoints": [ [ 450, 120 ], [ 460, 210 ], [ 430, 350 ], [ 350, 420 ], [ 460, 520 ], [ 480, 630 ], [ 580, 470 ] ], "backgroundPoints": [ [ 50, 350 ], [ 800, 350 ], [ 900, 100 ], [ 100, 600 ], [ 700, 600 ], [ 200, 200 ] ], "minimumProbeAccuracy": 0.9 }, "source": "WebAI independently annotated whole-person foreground/background probes on the Pexels portrait, visually reviewed at original 1024x683 resolution.", "license": "Pexels License", "id": "person-box", "userInput": { "input_boxes": [ [ 230, 40, 740, 683 ] ] } }, { "file": "model-samples/desktop/mobilenet-dog.jpg", "expected": { "minimumForegroundRatio": 0.25, "maximumForegroundRatio": 0.8, "foregroundPoints": [ [ 200, 150 ], [ 350, 200 ], [ 500, 400 ] ], "backgroundPoints": [ [ 0, 0 ], [ 690, 570 ], [ 680, 10 ] ], "minimumProbeAccuracy": 0.9 }, "source": "https://commons.wikimedia.org/wiki/File:YellowLabradorLooking_new.jpg — Elf; derivative Djmirko/FT2; independent visual foreground/background probes.", "license": "CC-BY-SA-3.0", "id": "dog-box", "userInput": { "input_boxes": [ [ 50, 30, 655, 560 ] ] } } ], "familyId": "mvanet", "runtime": "Transformers.js4.3.2 preferred browser; pinned native Transformers architecture evaluated before ONNX/PyTorch alternatives", "preprocess": "Pinned processor bilinear1024/ImageNet normalization; exact graph signatures determine sigmoid handling. No BEN2 timing reuse." }
Tested
2026-11-16T13:22:34Z
Environment
Apple M3 Max macOS27.2 ONNXRuntime1.23.2 FP32 four threads; existing mirror, product dev/build active. CoreML partial GPU profiled with actual per-operation compute-plan evidence.
Runtime
onnxruntime 1.23.2
Artifact revision
43ec3427514b8d9164eea02df93ca2f1b036bb7b
Fixture
{
  "file": "model-samples/modnet-portrait-pexels-5965592.jpg",
  "size": 78124,
  "sha1": "38212739a4a4f0be25d84863cf991d7fb6c6d818",
  "sha256": "2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe",
  "url": "https://assets.localai.tools/samples/v1/files/2b5d6423b8948eb53319661633039912afa6455a9055f0c984ff88eeae1b5bfe/modnet-portrait-pexels-5965592.jpg"
}
Rerun when
artifact revision; runtime version; fixture checksum; quality gate; backend behavior
Per-case quality & failure details
[
  {
    "precision": "fp32",
    "device": "coreml",
    "status": "pass",
    "quality": {
      "platform": "native-coreml",
      "id": "portrait",
      "source": "research/mvanet/native-coreml-portrait-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.26036042734260617,
      "probes": [
        {
          "point": [
            450,
            120
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            210
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            430,
            350
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            420
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            520
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            480,
            630
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            580,
            470
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            50,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            800,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            900,
            100
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            100,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            700,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            200,
            200
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true,
      "modnetReferenceAgreement": {
        "iou": 0.9787453502024513,
        "alphaMAE": 0.00637594692891798,
        "note": "Model-reference agreement, not human ground truth"
      }
    },
    "runtime": {
      "inferenceMs": 2185.501915999339,
      "durationMs": 2265.6409580013133,
      "imagesPerSecond": 0.4575607976727587,
      "placement": {
        "CPUExecutionProvider": 368,
        "CoreMLExecutionProvider": 105
      },
      "gpuPlanOperations": 2041,
      "gpuPlanSamples": [
        "ios18.concat, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.005081",
        "ios18.conv, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.056644",
        "ios18.reshape, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.014845",
        "ios16.reduce_mean, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.007481",
        "ios18.sub, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.002717"
      ],
      "provider": "CoreMLExecutionProvider",
      "acceleration": "partial",
      "visualizationMs": 2.3765420010022353,
      "loadMs": 156652.0287500025,
      "totalMs": 158837.53066600184,
      "grade": "good",
      "actualDevice": "coreml",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "coreml",
    "status": "pass",
    "quality": {
      "platform": "native-coreml",
      "id": "portrait",
      "source": "research/mvanet/native-coreml-portrait-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.26036042734260617,
      "probes": [
        {
          "point": [
            450,
            120
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            210
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            430,
            350
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            420
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            520
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            480,
            630
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            580,
            470
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            50,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            800,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            900,
            100
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            100,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            700,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            200,
            200
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true,
      "modnetReferenceAgreement": {
        "iou": 0.9787453502024513,
        "alphaMAE": 0.00637594692891798,
        "note": "Model-reference agreement, not human ground truth"
      }
    },
    "runtime": {
      "inferenceMs": 1568.2138750016748,
      "durationMs": 1638.3296670028358,
      "imagesPerSecond": 0.637668124189331,
      "placement": {
        "CPUExecutionProvider": 368,
        "CoreMLExecutionProvider": 105
      },
      "gpuPlanOperations": 2041,
      "gpuPlanSamples": [
        "ios18.concat, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.005081",
        "ios18.conv, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.056644",
        "ios18.reshape, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.014845",
        "ios16.reduce_mean, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.007481",
        "ios18.sub, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.002717"
      ],
      "provider": "CoreMLExecutionProvider",
      "acceleration": "partial",
      "visualizationMs": 0,
      "loadMs": 156652.0287500025,
      "totalMs": 158220.24262500418,
      "grade": "good",
      "actualDevice": "coreml",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "coreml",
    "status": "pass",
    "quality": {
      "platform": "native-coreml",
      "id": "portrait",
      "source": "research/mvanet/native-coreml-portrait-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.26036042734260617,
      "probes": [
        {
          "point": [
            450,
            120
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            210
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            430,
            350
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            420
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            520
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            480,
            630
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            580,
            470
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            50,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            800,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            900,
            100
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            100,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            700,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            200,
            200
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true,
      "modnetReferenceAgreement": {
        "iou": 0.9787453502024513,
        "alphaMAE": 0.00637594692891798,
        "note": "Model-reference agreement, not human ground truth"
      }
    },
    "runtime": {
      "inferenceMs": 1425.53683299775,
      "durationMs": 1491.8792080025014,
      "imagesPerSecond": 0.7014901171631658,
      "placement": {
        "CPUExecutionProvider": 368,
        "CoreMLExecutionProvider": 105
      },
      "gpuPlanOperations": 2041,
      "gpuPlanSamples": [
        "ios18.concat, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.005081",
        "ios18.conv, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.056644",
        "ios18.reshape, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.014845",
        "ios16.reduce_mean, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.007481",
        "ios18.sub, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.002717"
      ],
      "provider": "CoreMLExecutionProvider",
      "acceleration": "partial",
      "visualizationMs": 0,
      "loadMs": 156652.0287500025,
      "totalMs": 158077.56558300025,
      "grade": "good",
      "actualDevice": "coreml",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "coreml",
    "status": "pass",
    "quality": {
      "platform": "native-coreml",
      "id": "dog",
      "source": "research/mvanet/native-coreml-dog-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.3705595444416935,
      "probes": [
        {
          "point": [
            200,
            150
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            200
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            500,
            400
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            0,
            0
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            690,
            570
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            680,
            10
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true
    },
    "runtime": {
      "inferenceMs": 1338.960708999366,
      "durationMs": 1400.6762919998437,
      "imagesPerSecond": 0.746847904706122,
      "placement": {
        "CPUExecutionProvider": 368,
        "CoreMLExecutionProvider": 105
      },
      "gpuPlanOperations": 2041,
      "gpuPlanSamples": [
        "ios18.concat, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.005081",
        "ios18.conv, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.056644",
        "ios18.reshape, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.014845",
        "ios16.reduce_mean, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.007481",
        "ios18.sub, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.002717"
      ],
      "provider": "CoreMLExecutionProvider",
      "acceleration": "partial",
      "visualizationMs": 4.228457997669466,
      "loadMs": 156652.0287500025,
      "totalMs": 157990.98945900187,
      "grade": "good",
      "actualDevice": "coreml",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "coreml",
    "status": "pass",
    "quality": {
      "platform": "native-coreml",
      "id": "dog",
      "source": "research/mvanet/native-coreml-dog-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.3705595444416935,
      "probes": [
        {
          "point": [
            200,
            150
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            200
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            500,
            400
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            0,
            0
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            690,
            570
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            680,
            10
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true
    },
    "runtime": {
      "inferenceMs": 1443.6210420026327,
      "durationMs": 1496.2221250025323,
      "imagesPerSecond": 0.6927025659121532,
      "placement": {
        "CPUExecutionProvider": 368,
        "CoreMLExecutionProvider": 105
      },
      "gpuPlanOperations": 2041,
      "gpuPlanSamples": [
        "ios18.concat, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.005081",
        "ios18.conv, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.056644",
        "ios18.reshape, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.014845",
        "ios16.reduce_mean, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.007481",
        "ios18.sub, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.002717"
      ],
      "provider": "CoreMLExecutionProvider",
      "acceleration": "partial",
      "visualizationMs": 0,
      "loadMs": 156652.0287500025,
      "totalMs": 158095.64979200513,
      "grade": "good",
      "actualDevice": "coreml",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "coreml",
    "status": "pass",
    "quality": {
      "platform": "native-coreml",
      "id": "dog",
      "source": "research/mvanet/native-coreml-dog-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.3705595444416935,
      "probes": [
        {
          "point": [
            200,
            150
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            200
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            500,
            400
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            0,
            0
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            690,
            570
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            680,
            10
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true
    },
    "runtime": {
      "inferenceMs": 4485.236167001858,
      "durationMs": 4586.784999999509,
      "imagesPerSecond": 0.22295370026601002,
      "placement": {
        "CPUExecutionProvider": 368,
        "CoreMLExecutionProvider": 105
      },
      "gpuPlanOperations": 2041,
      "gpuPlanSamples": [
        "ios18.concat, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.005081",
        "ios18.conv, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.056644",
        "ios18.reshape, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.014845",
        "ios16.reduce_mean, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.007481",
        "ios18.sub, Device Usage: <MLGPUComputeDevice: 0x7859659570> Apple M3 Max, Estimated Cost: 0.002717"
      ],
      "provider": "CoreMLExecutionProvider",
      "acceleration": "partial",
      "visualizationMs": 0,
      "loadMs": 156652.0287500025,
      "totalMs": 161137.26491700436,
      "grade": "good",
      "actualDevice": "coreml",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "platform": "native-cpu",
      "id": "portrait",
      "source": "research/mvanet/native-cpu-portrait-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.2588033606332357,
      "probes": [
        {
          "point": [
            450,
            120
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            210
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            430,
            350
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            420
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            520
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            480,
            630
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            580,
            470
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            50,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            800,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            900,
            100
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            100,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            700,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            200,
            200
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true,
      "modnetReferenceAgreement": {
        "iou": 0.9848693108111838,
        "alphaMAE": 0.004799440365458043,
        "note": "Model-reference agreement, not human ground truth"
      }
    },
    "runtime": {
      "inferenceMs": 5821.9347080012085,
      "durationMs": 5914.803834002669,
      "imagesPerSecond": 0.17176420728760128,
      "placement": {
        "CPUExecutionProvider": 1708
      },
      "gpuPlanOperations": 0,
      "gpuPlanSamples": [],
      "provider": "CPUExecutionProvider",
      "acceleration": "cpu",
      "visualizationMs": 2.379417001066031,
      "loadMs": 344.3396250004298,
      "totalMs": 6166.274333001638,
      "grade": "acceptable",
      "actualDevice": "cpu",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "platform": "native-cpu",
      "id": "portrait",
      "source": "research/mvanet/native-cpu-portrait-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.2588033606332357,
      "probes": [
        {
          "point": [
            450,
            120
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            210
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            430,
            350
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            420
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            520
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            480,
            630
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            580,
            470
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            50,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            800,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            900,
            100
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            100,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            700,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            200,
            200
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true,
      "modnetReferenceAgreement": {
        "iou": 0.9848693108111838,
        "alphaMAE": 0.004799440365458043,
        "note": "Model-reference agreement, not human ground truth"
      }
    },
    "runtime": {
      "inferenceMs": 5660.323000000062,
      "durationMs": 5723.957583999436,
      "imagesPerSecond": 0.17666836327184668,
      "placement": {
        "CPUExecutionProvider": 1708
      },
      "gpuPlanOperations": 0,
      "gpuPlanSamples": [],
      "provider": "CPUExecutionProvider",
      "acceleration": "cpu",
      "visualizationMs": 0,
      "loadMs": 344.3396250004298,
      "totalMs": 6004.662625000492,
      "grade": "acceptable",
      "actualDevice": "cpu",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "platform": "native-cpu",
      "id": "portrait",
      "source": "research/mvanet/native-cpu-portrait-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.2588033606332357,
      "probes": [
        {
          "point": [
            450,
            120
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            210
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            430,
            350
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            420
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            460,
            520
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            480,
            630
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            580,
            470
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            50,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            800,
            350
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            900,
            100
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            100,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            700,
            600
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            200,
            200
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true,
      "modnetReferenceAgreement": {
        "iou": 0.9848693108111838,
        "alphaMAE": 0.004799440365458043,
        "note": "Model-reference agreement, not human ground truth"
      }
    },
    "runtime": {
      "inferenceMs": 5662.1741660019325,
      "durationMs": 5731.118583000352,
      "imagesPerSecond": 0.1766106041040594,
      "placement": {
        "CPUExecutionProvider": 1708
      },
      "gpuPlanOperations": 0,
      "gpuPlanSamples": [],
      "provider": "CPUExecutionProvider",
      "acceleration": "cpu",
      "visualizationMs": 0,
      "loadMs": 344.3396250004298,
      "totalMs": 6006.513791002362,
      "grade": "acceptable",
      "actualDevice": "cpu",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "platform": "native-cpu",
      "id": "dog",
      "source": "research/mvanet/native-cpu-dog-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.37355038375835603,
      "probes": [
        {
          "point": [
            200,
            150
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            200
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            500,
            400
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            0,
            0
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            690,
            570
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            680,
            10
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true
    },
    "runtime": {
      "inferenceMs": 5385.083125001984,
      "durationMs": 5440.623917002085,
      "imagesPerSecond": 0.18569815484503438,
      "placement": {
        "CPUExecutionProvider": 1708
      },
      "gpuPlanOperations": 0,
      "gpuPlanSamples": [],
      "provider": "CPUExecutionProvider",
      "acceleration": "cpu",
      "visualizationMs": 2.6658749993657693,
      "loadMs": 344.3396250004298,
      "totalMs": 5729.4227500024135,
      "grade": "acceptable",
      "actualDevice": "cpu",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "platform": "native-cpu",
      "id": "dog",
      "source": "research/mvanet/native-cpu-dog-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.37355038375835603,
      "probes": [
        {
          "point": [
            200,
            150
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            200
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            500,
            400
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            0,
            0
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            690,
            570
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            680,
            10
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true
    },
    "runtime": {
      "inferenceMs": 5446.270791999268,
      "durationMs": 5499.528124997596,
      "imagesPerSecond": 0.18361187649153057,
      "placement": {
        "CPUExecutionProvider": 1708
      },
      "gpuPlanOperations": 0,
      "gpuPlanSamples": [],
      "provider": "CPUExecutionProvider",
      "acceleration": "cpu",
      "visualizationMs": 0,
      "loadMs": 344.3396250004298,
      "totalMs": 5790.610416999698,
      "grade": "acceptable",
      "actualDevice": "cpu",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  },
  {
    "precision": "fp32",
    "device": "cpu",
    "status": "pass",
    "quality": {
      "platform": "native-cpu",
      "id": "dog",
      "source": "research/mvanet/native-cpu-dog-0-output.json",
      "probeAccuracy": 1,
      "foregroundRatio": 0.37355038375835603,
      "probes": [
        {
          "point": [
            200,
            150
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            350,
            200
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            500,
            400
          ],
          "foreground": true,
          "passed": true
        },
        {
          "point": [
            0,
            0
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            690,
            570
          ],
          "foreground": false,
          "passed": true
        },
        {
          "point": [
            680,
            10
          ],
          "foreground": false,
          "passed": true
        }
      ],
      "passed": true
    },
    "runtime": {
      "inferenceMs": 5424.098708001111,
      "durationMs": 5476.016999997228,
      "imagesPerSecond": 0.1843624266138254,
      "placement": {
        "CPUExecutionProvider": 1708
      },
      "gpuPlanOperations": 0,
      "gpuPlanSamples": [],
      "provider": "CPUExecutionProvider",
      "acceleration": "cpu",
      "visualizationMs": 0,
      "loadMs": 344.3396250004298,
      "totalMs": 5768.438333001541,
      "grade": "acceptable",
      "actualDevice": "cpu",
      "actualPrecision": "fp32",
      "engine": "onnxruntime",
      "version": "1.23.2",
      "artifactRevision": "43ec3427514b8d9164eea02df93ca2f1b036bb7b"
    }
  }
]
Release validation still required
  • Complete verified execution dependency backup and forced cold model/runtime source recovery acceptance.
  • Current packaged app acceptance is tracked separately in research/mvanet/PROGRESS.md.

Something not quite right?

Tell us what happened with MVANet Desktop. Loading, results, or the page itself—we’re here for all of it.

Attachments Optional

Up to 3 files · 50 MB each

Browser, OS, CPU/GPU, memory, screen, storage, and network details are included automatically. Model inputs, results, and browsing history are not collected.