EfficientSAM-S INT8

Cloud CPUAWS Graviton G4

INT8-quantized box-prompted image segmentation model with a ViT-Small backbone and lightweight mask decoder. It accepts a 1024×1024 RGB image plus bounding-box coordinates and returns a binary mask for the prompted region, offering reduced model size and faster inference while preserving segmentation quality.View on Hugging Face

Accuracyvs base model

mIoU74.78%↓0 pp

Mask mAP@0.50:0.9541.92%↓0.01 pp

Frames / sec

About this version

Original Model:: efficient_sam_s by Meta — efficient_sam_s
Optimization/conversion:: Arm-Optimized version for execution on Arm-based platforms.
Converted/optimized by:: Arm
License:: The Original Model and the Optimized Model are subject to Apache-2.0.

This repository contains a converted or optimized version of the Original Model (the “Optimized Model”). The Original Model has been converted or optimized as described above for execution on Arm-based platforms.

No retraining or fine-tuning of the Original Model was performed as part of the conversion or optimization. The conversion or optimization was not intended to change the Original Model’s behavior or intended use.

Original Model and Documentation

For information about the Original Model, including its development, training data, intended uses, limitations and other relevant information, please refer to the Original Model repository. Information in that repository was provided by the original developer or other third parties and, unless expressly stated otherwise, has not been independently verified by Arm.

Licenses and Third-Party Terms

Use of the Original Model and the Optimized Model is subject to the applicable licenses, usage restrictions and other terms identified above and in the relevant repositories. Publication of the Optimized Model does not grant any rights beyond those provided under the applicable license terms.

You are responsible for reviewing those terms and ensuring that your use of the Original Model and the Optimized Model is permitted.

Purpose of this Release

About this version

Original Model:: efficient_sam_s by Meta — efficient_sam_s
Optimization/conversion:: Arm-Optimized version for execution on Arm-based platforms.
Converted/optimized by:: Arm
License:: The Original Model and the Optimized Model are subject to Apache-2.0.

Original Model and Documentation

Licenses and Third-Party Terms

You are responsible for reviewing those terms and ensuring that your use of the Original Model and the Optimized Model is permitted.

Metric	Resultsvs baseline	Change
End to end latency p50	1573.85ms2030.93ms	−457.08 ms
End to end latency p90	1617.13ms2092.38ms	−475.24 ms
End to end latency p99	1646.36ms2108.40ms	−462.05 ms
Peak memory	913.91MB1655.73MB	−741.82 MB
Average memory	911.58MB1351.86MB	−440.28 MB
Time to first inference	1640.65ms2247.35ms	−606.69 ms
Frames per second	0.64fps0.49fps	+0.15 fps
mIoU	74.78%74.78%	0.00%
Mask mAP@0.50	68.87%68.86%	+0.01%
Mask mAP@0.75	43.97%44.00%	−0.04%
Mask mAP@0.50:0.95	41.92%41.94%	−0.01%

Configuration	Baseline	Arm Optimized
Dataset	COCO 2017 (val2017)	COCO 2017 (val2017)
Sample count	5,000	5,000
Batch size	1	1
Input resolution	1024x1024	1024x1024
Runs	50	50
Warmup runs	10	10

EfficientSAM-S INT8

EfficientSAM-S INT8

Model Evaluation

Configuration

Comparability