132 lines
3.8 KiB
Python
132 lines
3.8 KiB
Python
|
# -*- coding: utf-8 -*-
|
|||
|
|
"""
|
|||
|
|
AI Vision 模块 - CLIP Vision Encoder ONNX 推理
|
|||
|
|
"""
|
|||
|
|
|
|||
|
|
import os
|
|||
|
|
import numpy as np
|
|||
|
|
from PIL import Image
|
|||
|
|
import onnxruntime as ort
|
|||
|
|
|
|||
|
|
# ============================================================================
|
|||
|
|
# 全局模型单例（项目启动时加载一次）
|
|||
|
|
# ============================================================================
|
|||
|
|
|
|||
|
|
MODEL_PATH = os.path.join(os.path.dirname(__file__), '..', '..', 'models', 'clip_vision.onnx')
|
|||
|
|
|
|||
|
|
# 加载选项：CPU 推理，禁用依赖库的启动开销
|
|||
|
|
_session_options = ort.SessionOptions()
|
|||
|
|
_session_options.graph_optimization_level = ort.GraphOptimizationLevel.ORT_ENABLE_ALL
|
|||
|
|
|
|||
|
|
ort_session: ort.InferenceSession = None
|
|||
|
|
|
|||
|
|
|
|||
|
|
def load_clip_model():
|
|||
|
|
"""启动时调用：全局加载 CLIP Vision 模型"""
|
|||
|
|
global ort_session
|
|||
|
|
if ort_session is not None:
|
|||
|
|
return ort_session
|
|||
|
|
|
|||
|
|
if not os.path.exists(MODEL_PATH):
|
|||
|
|
raise FileNotFoundError(f"CLIP Vision 模型未找到: {MODEL_PATH}")
|
|||
|
|
|
|||
|
|
ort_session = ort.InferenceSession(MODEL_PATH, sess_options=_session_options, providers=['CPUExecutionProvider'])
|
|||
|
|
print(f"✅ [AI Vision] CLIP 模型加载成功: {MODEL_PATH}")
|
|||
|
|
return ort_session
|
|||
|
|
|
|||
|
|
|
|||
|
|
# ============================================================================
|
|||
|
|
# CLIP 预处理常量
|
|||
|
|
# ============================================================================
|
|||
|
|
|
|||
|
|
# ImageNet 标准归一化（CLIP 官方）
|
|||
|
|
IMAGENET_MEAN = [0.485, 0.456, 0.406]
|
|||
|
|
IMAGENET_STD = [0.229, 0.224, 0.225]
|
|||
|
|
|
|||
|
|
# 模型输入尺寸
|
|||
|
|
INPUT_SIZE = 224
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _center_crop_and_resize(image: Image.Image) -> Image.Image:
|
|||
|
|
"""
|
|||
|
|
CLIP 官方预处理：中心裁剪抗干扰
|
|||
|
|
- 将图片最短边缩放到 224
|
|||
|
|
- 从正中间切取 224x224 区域
|
|||
|
|
"""
|
|||
|
|
w, h = image.size
|
|||
|
|
|
|||
|
|
# 计算缩放后的目标尺寸
|
|||
|
|
if w < h:
|
|||
|
|
new_w = INPUT_SIZE
|
|||
|
|
new_h = int(h * INPUT_SIZE / w)
|
|||
|
|
else:
|
|||
|
|
new_h = INPUT_SIZE
|
|||
|
|
new_w = int(w * INPUT_SIZE / h)
|
|||
|
|
|
|||
|
|
# 缩放
|
|||
|
|
image = image.resize((new_w, new_h), Image.BILINEAR)
|
|||
|
|
|
|||
|
|
# 中心裁剪
|
|||
|
|
left = (new_w - INPUT_SIZE) // 2
|
|||
|
|
top = (new_h - INPUT_SIZE) // 2
|
|||
|
|
right = left + INPUT_SIZE
|
|||
|
|
bottom = top + INPUT_SIZE
|
|||
|
|
|
|||
|
|
return image.crop((left, top, right, bottom))
|
|||
|
|
|
|||
|
|
|
|||
|
|
def _normalize(image_np: np.ndarray) -> np.ndarray:
|
|||
|
|
"""
|
|||
|
|
对 224x224x3 图像进行 CLIP 标准归一化
|
|||
|
|
image_np: shape (H, W, C), dtype uint8, 值域 [0, 255]
|
|||
|
|
返回: shape (C, H, W), dtype float32, 值域 [0, 1]
|
|||
|
|
"""
|
|||
|
|
# HWC -> CHW
|
|||
|
|
image_np = image_np.transpose(2, 0, 1).astype(np.float32) / 255.0
|
|||
|
|
|
|||
|
|
# 归一化
|
|||
|
|
for i, (mean, std) in enumerate(zip(IMAGENET_MEAN, IMAGENET_STD)):
|
|||
|
|
image_np[i] = (image_np[i] - mean) / std
|
|||
|
|
|
|||
|
|
return image_np
|
|||
|
|
|
|||
|
|
|
|||
|
|
# ============================================================================
|
|||
|
|
# 主函数：提取图像 embedding
|
|||
|
|
# ============================================================================
|
|||
|
|
|
|||
|
|
def get_image_embedding(image_path: str) -> list:
|
|||
|
|
"""
|
|||
|
|
提取图像的 512 维 CLIP embedding 向量
|
|||
|
|
|
|||
|
|
参数:
|
|||
|
|
image_path: 图像文件路径（支持本地路径或 URL）
|
|||
|
|
|
|||
|
|
返回:
|
|||
|
|
list: 512 维浮点向量
|
|||
|
|
"""
|
|||
|
|
if ort_session is None:
|
|||
|
|
load_clip_model()
|
|||
|
|
|
|||
|
|
# 加载图像
|
|||
|
|
try:
|
|||
|
|
image = Image.open(image_path).convert('RGB')
|
|||
|
|
except Exception as e:
|
|||
|
|
raise ValueError(f"图像加载失败: {image_path}, 错误: {e}")
|
|||
|
|
|
|||
|
|
# 中心裁剪
|
|||
|
|
image = _center_crop_and_resize(image)
|
|||
|
|
|
|||
|
|
# 归一化
|
|||
|
|
input_data = _normalize(np.array(image))
|
|||
|
|
|
|||
|
|
# 添加 batch 维度: (C, H, W) -> (1, C, H, W)
|
|||
|
|
input_data = np.expand_dims(input_data, axis=0)
|
|||
|
|
|
|||
|
|
# 推理
|
|||
|
|
outputs = ort_session.run(None, {'images': input_data.astype(np.float32)})
|
|||
|
|
|
|||
|
|
# 输出通常是 (1, 512) 的向量，取第一项并展平为 list
|
|||
|
|
embedding = outputs[0][0].tolist()
|
|||
|
|
|
|||
|
|
return embedding
|