import onnxruntime as ort
import numpy as np

sess = ort.InferenceSession("data/onnx/face_parsing.farl.celebm.main_ema_181500_jit.onnx", providers=["CPUExecutionProvider"])
input_name = sess.get_inputs()[0].name

dummy = np.zeros((1, 3, 448, 448), dtype=np.float32)
logits, aux = sess.run(None, {input_name: dummy})
print(logits.shape, aux.shape)

<!-- ### FaRL(face_parsing.farl.celebm.main_ema_181500_jit.onnx)

- 输入:NCHW,float32,B×3×448×448

预处理:RGB,0-1 归一化

- 输出:

- logits:B×C×512×512

CelebM 为 C=19,LaPa 为 C=11

- aux:B×1

说明:FaRL 输出为 512×512,需要在推理端将 logits resize 到 448×448 后再做反变换。 -->

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support