Re-submit: Fix: Proper RGBA -> RGB conversion for PIL images. (#18569)
Signed-off-by: Chenheli Hua <huachenheli@outlook.com>
This commit is contained in:
@@ -4,6 +4,7 @@ import pytest
|
||||
|
||||
from vllm.assets.image import ImageAsset
|
||||
from vllm.assets.video import VideoAsset
|
||||
from vllm.multimodal.image import convert_image_mode
|
||||
|
||||
models = ["llava-hf/llava-onevision-qwen2-0.5b-ov-hf"]
|
||||
|
||||
@@ -26,8 +27,9 @@ def test_models(vllm_runner, model, dtype: str, max_tokens: int) -> None:
|
||||
give the same result.
|
||||
"""
|
||||
|
||||
image_cherry = ImageAsset("cherry_blossom").pil_image.convert("RGB")
|
||||
image_stop = ImageAsset("stop_sign").pil_image.convert("RGB")
|
||||
image_cherry = convert_image_mode(
|
||||
ImageAsset("cherry_blossom").pil_image, "RGB")
|
||||
image_stop = convert_image_mode(ImageAsset("stop_sign").pil_image, "RGB")
|
||||
images = [image_cherry, image_stop]
|
||||
video = VideoAsset(name="baby_reading", num_frames=16).np_ndarrays
|
||||
|
||||
|
||||
@@ -12,7 +12,7 @@ from transformers import AutoTokenizer
|
||||
|
||||
from vllm.assets.image import ImageAsset
|
||||
from vllm.lora.request import LoRARequest
|
||||
from vllm.multimodal.image import rescale_image_size
|
||||
from vllm.multimodal.image import convert_image_mode, rescale_image_size
|
||||
from vllm.platforms import current_platform
|
||||
from vllm.sequence import SampleLogprobs
|
||||
|
||||
@@ -267,7 +267,7 @@ def test_vision_speech_models(hf_runner, vllm_runner, model, dtype: str,
|
||||
|
||||
# use the example speech question so that the model outputs are reasonable
|
||||
audio = librosa.load(speech_question, sr=None)
|
||||
image = ImageAsset("cherry_blossom").pil_image.convert("RGB")
|
||||
image = convert_image_mode(ImageAsset("cherry_blossom").pil_image, "RGB")
|
||||
|
||||
inputs_vision_speech = [
|
||||
(
|
||||
|
||||
@@ -4,6 +4,7 @@ import pytest
|
||||
|
||||
from vllm import LLM, SamplingParams
|
||||
from vllm.assets.image import ImageAsset
|
||||
from vllm.multimodal.image import convert_image_mode
|
||||
|
||||
from ..utils import create_new_process_for_each_test
|
||||
|
||||
@@ -58,7 +59,7 @@ def test_oot_registration_embedding(
|
||||
assert all(v == 0 for v in output.outputs.embedding)
|
||||
|
||||
|
||||
image = ImageAsset("cherry_blossom").pil_image.convert("RGB")
|
||||
image = convert_image_mode(ImageAsset("cherry_blossom").pil_image, "RGB")
|
||||
|
||||
|
||||
@create_new_process_for_each_test()
|
||||
|
||||
Reference in New Issue
Block a user