BiSeNet face parsing (fastforks)
BiSeNet (Yu et al. 2018) face parsing from face-parsing.PyTorch: a ResNet-18 context path with attention refinement and feature fusion, 19 CelebAMask-HQ face-part classes (skin, brows, eyes, glasses, ears, earrings, nose, mouth, lips, neck, necklace, cloth, hair, hat) on a 512 x 512 face crop. The parser that GFPGAN / CodeFormer pipelines use through facexlib; MIT code and weights.
This repo is the model card only: it holds no weights. fastforks downloads the authors' weights from the links below and checks them against the SHA-256 in the model card. Code: https://github.com/kleinicke/fastforks. Kernels for the kernels backend: https://huggingface.co/fastforks/kernels. Original code: https://github.com/xinntao/facexlib at 260620ae9399.
Task: segmentation. Tier: experimental. Outputs: logits. Backends: portable.
Input: one RGB image. Output: per-pixel class labels (logits with logits=True).
Variants
Weights are the authors' files (not hosted here). SHA-256 shortened to 12 hex digits; the full hashes are in card.toml.
| variant | weights (author link) | SHA-256 | weights licence | commercial use |
|---|---|---|---|---|
bisenet-face-parsing@1 |
parsing_bisenet.pth | 468e13ca13a9 |
MIT | yes |
Licence
Original-code strings, recorded as they are (research-use package; check them before any other use). Code: MIT. Licence files: LICENSE.
Training data: ImageNet-1k (ILSVRC 2012 classification), CelebAMask-HQ (30 000 CelebA-HQ faces, 19 face-part masks) (face-parsing.PyTorch train.py: CelebAMask-HQ (non-commercial research only), ResNet-18 initialised from torchvision's ImageNet weights.).
Licence evidence: https://github.com/zllrunning/face-parsing.PyTorch/blob/d2e684cf1588b46145635e8fe7bcc29544e5537e/LICENSE ; https://github.com/xinntao/facexlib/blob/260620ae93990a300f4b16448df9bb459f1caba9/LICENSE ; https://github.com/xinntao/facexlib/blob/260620ae93990a300f4b16448df9bb459f1caba9/facexlib/parsing/__init__.py
Input size
any H x W >= 1, resized or padded inside the model
Tested up to (portable, fp16, RTX 5060 Ti 16 GB): bisenet-face-parsing 512x512 (0.0 GiB peak allocated, 0.1 reserved)
Speed and memory
Segmentation, synthetic
| model | input | time (ms): original fp32 | original | portable | kernels | kernels speed-up: vs original | vs portable | peak memory (MiB, allocated / reserved): original | portable | kernels |
|---|---|---|---|---|---|---|---|---|---|---|
bisenet-face-parsing |
512x512 | β | 5.94 | 2.41 | β | β | β | 69 / 94 | 51 / 72 | β |
Startup and first call
| model | load (s): original | portable | kernels | first call (s): original | portable | kernels |
|---|---|---|---|---|---|---|
bisenet-face-parsing |
0.31 (512x512) | 0.20 (512x512) | β | 0.41 (512x512) | 0.28 (512x512) | β |
torch.compile and TensorRT
| model | input | time: portable | compile | TensorRT fp16 | kernels | preparation: compile (cold first call) | ONNX + engine build | peak MiB (alloc / reserved): portable | compile | kernels | TensorRT (engine + torch) |
|---|---|---|---|---|---|---|---|---|---|---|---|
bisenet-face-parsing |
512x512 | 2.41 ms | 1.34 ms | 0.678 ms | β | 5.79 s | 12.9 s | 51 / 72 | 76 / 110 | β | 85 / 110 |
Fixed-size cells: time (ms) / peak allocated / peak reserved memory (MiB).
Hardware, protocol and caveats: benchmarks (RTX 5060 Ti 16 GB, sm_120). Original = the original code named above at the precision of portable (fp16, or (fp32) where fp16 is unsupported or not finite); original fp32 = its default fp32 run. Peak memory is the torch allocator's.
Accuracy
Parity with the original code
Every variant is checked against reference outputs of the original code at its pinned commit. CPU, float32: max |diff| <= 2.6e-05 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.0042 and p99 <= 0.02 (absolute error of the output logits).
Usage
from fastforks import segmentation
model = segmentation.load("bisenet-face-parsing") # mode="auto": the preferred backend from recorded measurements
result = model("photo.jpg")
result.labels # [H, W] per-pixel class index
result.backend # what ran, e.g. Backend(mode='portable', device='cuda:0', dtype='float16')
Install with pip install fastforks (see https://github.com/kleinicke/fastforks). Other variants: bisenet-face-parsing.
Kernels backend
No kernels backend: measured slower: cudnn_hacc fp16-accumulate convolutions 2.77 ms vs 2.41 ms portable (512x512).