test_modeling_schp.py

from pathlib import Path

import pytest
import torch
from PIL import Image
from transformers import AutoImageProcessor, AutoModelForSemanticSegmentation

MODEL_PATH = str(Path(__file__).parent)


@pytest.fixture(scope="module")
def model():
    return AutoModelForSemanticSegmentation.from_pretrained(
        MODEL_PATH, trust_remote_code=True
    )


@pytest.fixture(scope="module")
def processor():
    return AutoImageProcessor.from_pretrained(MODEL_PATH, trust_remote_code=True)


def test_forward_shape(model):
    pixel_values = torch.randn(1, 3, 512, 512)
    with torch.no_grad():
        outputs = model(pixel_values=pixel_values)
    assert outputs.logits.shape == (1, 18, 512, 512)
    assert outputs.parsing_logits.shape == (1, 18, 512, 512)
    assert outputs.edge_logits.shape == (1, 2, 512, 512)


def test_seg_map(model):
    pixel_values = torch.randn(1, 3, 512, 512)
    with torch.no_grad():
        outputs = model(pixel_values=pixel_values)
    seg_map = outputs.logits.argmax(dim=1).squeeze()
    assert seg_map.shape == (512, 512)
    assert seg_map.min() >= 0
    assert seg_map.max() < 18


def test_processor_output_shape(processor):
    image = Image.new("RGB", (640, 480))
    inputs = processor(images=image, return_tensors="pt")
    assert "pixel_values" in inputs
    assert inputs["pixel_values"].shape == (1, 3, 512, 512)