File size: 1,791 Bytes
0515c22 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 |
import base64
import glob
from openai import OpenAI
import cv2
BASE_URL = "http://localhost:8000/v1"
def img_to_data_url(img_path: str):
img = cv2.imread(img_path)
if img is None:
raise FileNotFoundError(f"Cannot read image: {img_path}")
ok, buf = cv2.imencode(".jpg", img)
if not ok:
raise RuntimeError("cv2.imencode failed")
b64 = base64.b64encode(buf).decode("ascii")
return f"data:image/jpeg;base64,{b64}"
def test(openai_messages):
client = OpenAI(api_key="not-needed", base_url=BASE_URL)
stream = client.chat.completions.create(
model="AXERA-TECH/SmolVLM2-256M-Video-Instruct",
messages=openai_messages,
stream=True,
)
out_chunks = []
for ev in stream:
delta = ev.choices[0].delta
if delta and delta.content:
out_chunks.append(delta.content)
print(delta.content, end="", flush=True)
print()
assistant_text = "".join(out_chunks).strip()
def test_image():
image_data = img_to_data_url("video/frame_0000.jpg")
openai_messages = {
"role": "user",
"content": [
{"type": "text", "text": "Describe this image"},
{"type": "image_url", "image_url": image_data},
],
}
test(openai_messages)
def test_video():
image_list = glob.glob("video/*.jpg")
image_list.sort()
image_data_list = [img_to_data_url(img) for img in image_list]
openai_messages = {
"role": "user",
"content": [
{"type": "text", "text": "Describe this video"},
{"type": "image_url", "is_video":True, "image_url": image_data_list},
],
}
test(openai_messages)
print("Test image")
test_image()
print("Test video")
test_video()
|