README.md

metadata

license: apache-2.0
tags:
  - executorch
  - xnnpack
  - pte
  - on-device
  - image-segmentation
  - portrait-matting

modnet_portrait_matting — ExecuTorch

• Source: ZHKKKe/MODNet + DavG25/modnet-pretrained-models ckpt
• License: Apache-2.0
• Input: [[1, 3, 512, 512]] — RGB [-1,1], 512x512
• Output: alpha matte [1,1,512,512] 0-1

Variants

All variants take and return fp32 tensors — swap the .pte file, keep your app code.

build	file	size (MB)	parity vs fp32 eager (worst corr)	Mac median (ms)*
fp32	modnet_portrait_matting_xnnpack_fp32.pte	26.1	1.000000	64.9
fp16	modnet_portrait_matting_xnnpack_fp16.pte	24.4	1.000000	107.8
Core ML (fp16, iOS)	modnet_portrait_matting_coreml_all.pte	13.8	0.999997	6.8

The Core ML build is the same graph lowered to Apple's Neural Engine instead of XNNPACK, which is CPU-only. Measured on an iPhone 17 Pro across seven models, it runs 3.5x to 13.9x faster (median 12x) at roughly half the file size — for example Depth-Anything-V2-Small at 500.8 ms against 42.7 ms, and MODNet at 81.7 ms against 5.9 ms. It computes in fp16 and is iOS-only; the XNNPACK files stay the portable option and are what runs on Android.

*Mac arm64, single process, median of 10 — a reference point for relative cost only, not a device number (torch eager fp32 on the same machine: 120.1 ms).

Builds that did not earn a slot

• int8 is not shipped: measured in the units that matter for this model — mask IoU at 0.5: median 0.9864 over 10 real images, worst 0.5970.

Verification (executorch 1.4.0, torch 2.13.0)

Parity is measured against the fp32 eager model on real image input; corr is the correlation over all elements of each output tensor.

output	shape	max_abs_diff	corr
0	[1, 1, 512, 512]	1.476e-04	1.000000

XNNPACK delegate coverage (fp32): 94.7% (302/319 ops); ops left on the portable kernels: aten._native_batch_norm_legit.no_stats x16, aten.expand_copy.default x1

Conversion

torch.export -> to_edge_transform_and_lower(partitioner) -> .pte (conversion scripts: executorch-models)