File size: 10,753 Bytes
3079197 484e5ab 3079197 29b8637 79ada0b 41c7a59 d0db329 8f9784a 41c7a59 d0db329 436c3b5 d0db329 41c7a59 d0db329 3079197 d0db329 34b2ab3 d0db329 41c7a59 79ada0b 41c7a59 d0db329 e06e08c 3079197 41c7a59 d0db329 f5f274f e4c23fc d0db329 3079197 f5f274f d0db329 e32ef75 d0db329 29b8637 d0db329 eba9007 d0db329 e06e08c 3079197 41c7a59 79ada0b 41c7a59 79ada0b 41c7a59 3079197 d0db329 3079197 41c7a59 d0db329 41c7a59 e32ef75 5e0a689 e06e08c 5e0a689 41c7a59 5e0a689 1550520 8f9784a 6b3ce5a 63df91a 6b3ce5a 63df91a 1f5bc27 436c3b5 1f5bc27 8f9784a 436c3b5 1550520 e06e08c 1550520 |
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 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 |
#
# Copyright 2024 The InfiniFlow Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
from openai.lib.azure import AzureOpenAI
from zhipuai import ZhipuAI
import io
from abc import ABC
from ollama import Client
from PIL import Image
from openai import OpenAI
import os
import base64
from io import BytesIO
import json
import requests
from api.utils import get_uuid
from api.utils.file_utils import get_project_base_directory
class Base(ABC):
def __init__(self, key, model_name):
pass
def describe(self, image, max_tokens=300):
raise NotImplementedError("Please implement encode method!")
def image2base64(self, image):
if isinstance(image, bytes):
return base64.b64encode(image).decode("utf-8")
if isinstance(image, BytesIO):
return base64.b64encode(image.getvalue()).decode("utf-8")
buffered = BytesIO()
try:
image.save(buffered, format="JPEG")
except Exception as e:
image.save(buffered, format="PNG")
return base64.b64encode(buffered.getvalue()).decode("utf-8")
def prompt(self, b64):
return [
{
"role": "user",
"content": [
{
"type": "image_url",
"image_url": {
"url": f"data:image/jpeg;base64,{b64}"
},
},
{
"text": "请用中文详细描述一下图中的内容，比如时间，地点，人物，事情，人物心情等，如果有数据请提取出数据。" if self.lang.lower() == "chinese" else
"Please describe the content of this picture, like where, when, who, what happen. If it has number data, please extract them out.",
},
],
}
]
class GptV4(Base):
def __init__(self, key, model_name="gpt-4-vision-preview", lang="Chinese", base_url="https://api.openai.com/v1"):
if not base_url: base_url="https://api.openai.com/v1"
self.client = OpenAI(api_key=key, base_url=base_url)
self.model_name = model_name
self.lang = lang
def describe(self, image, max_tokens=300):
b64 = self.image2base64(image)
prompt = self.prompt(b64)
for i in range(len(prompt)):
for c in prompt[i]["content"]:
if "text" in c: c["type"] = "text"
res = self.client.chat.completions.create(
model=self.model_name,
messages=prompt,
max_tokens=max_tokens,
)
return res.choices[0].message.content.strip(), res.usage.total_tokens
class AzureGptV4(Base):
def __init__(self, key, model_name, lang="Chinese", **kwargs):
self.client = AzureOpenAI(api_key=key, azure_endpoint=kwargs["base_url"], api_version="2024-02-01")
self.model_name = model_name
self.lang = lang
def describe(self, image, max_tokens=300):
b64 = self.image2base64(image)
prompt = self.prompt(b64)
for i in range(len(prompt)):
for c in prompt[i]["content"]:
if "text" in c: c["type"] = "text"
res = self.client.chat.completions.create(
model=self.model_name,
messages=prompt,
max_tokens=max_tokens,
)
return res.choices[0].message.content.strip(), res.usage.total_tokens
class QWenCV(Base):
def __init__(self, key, model_name="qwen-vl-chat-v1", lang="Chinese", **kwargs):
import dashscope
dashscope.api_key = key
self.model_name = model_name
self.lang = lang
def prompt(self, binary):
# stupid as hell
tmp_dir = get_project_base_directory("tmp")
if not os.path.exists(tmp_dir):
os.mkdir(tmp_dir)
path = os.path.join(tmp_dir, "%s.jpg" % get_uuid())
Image.open(io.BytesIO(binary)).save(path)
return [
{
"role": "user",
"content": [
{
"image": f"file://{path}"
},
{
"text": "请用中文详细描述一下图中的内容，比如时间，地点，人物，事情，人物心情等，如果有数据请提取出数据。" if self.lang.lower() == "chinese" else
"Please describe the content of this picture, like where, when, who, what happen. If it has number data, please extract them out.",
},
],
}
]
def describe(self, image, max_tokens=300):
from http import HTTPStatus
from dashscope import MultiModalConversation
response = MultiModalConversation.call(model=self.model_name,
messages=self.prompt(image))
if response.status_code == HTTPStatus.OK:
return response.output.choices[0]['message']['content'][0]["text"], response.usage.output_tokens
return response.message, 0
class Zhipu4V(Base):
def __init__(self, key, model_name="glm-4v", lang="Chinese", **kwargs):
self.client = ZhipuAI(api_key=key)
self.model_name = model_name
self.lang = lang
def describe(self, image, max_tokens=1024):
b64 = self.image2base64(image)
res = self.client.chat.completions.create(
model=self.model_name,
messages=self.prompt(b64),
max_tokens=max_tokens,
)
return res.choices[0].message.content.strip(), res.usage.total_tokens
class OllamaCV(Base):
def __init__(self, key, model_name, lang="Chinese", **kwargs):
self.client = Client(host=kwargs["base_url"])
self.model_name = model_name
self.lang = lang
def describe(self, image, max_tokens=1024):
prompt = self.prompt("")
try:
options = {"num_predict": max_tokens}
response = self.client.generate(
model=self.model_name,
prompt=prompt[0]["content"][1]["text"],
images=[image],
options=options
)
ans = response["response"].strip()
return ans, 128
except Exception as e:
return "**ERROR**: " + str(e), 0
class XinferenceCV(Base):
def __init__(self, key, model_name="", lang="Chinese", base_url=""):
self.client = OpenAI(api_key="xxx", base_url=base_url)
self.model_name = model_name
self.lang = lang
def describe(self, image, max_tokens=300):
b64 = self.image2base64(image)
res = self.client.chat.completions.create(
model=self.model_name,
messages=self.prompt(b64),
max_tokens=max_tokens,
)
return res.choices[0].message.content.strip(), res.usage.total_tokens
class GeminiCV(Base):
def __init__(self, key, model_name="gemini-1.0-pro-vision-latest", lang="Chinese", **kwargs):
from google.generativeai import client,GenerativeModel
client.configure(api_key=key)
_client = client.get_default_generative_client()
self.model_name = model_name
self.model = GenerativeModel(model_name=self.model_name)
self.model._client = _client
self.lang = lang
def describe(self, image, max_tokens=2048):
from PIL.Image import open
gen_config = {'max_output_tokens':max_tokens}
prompt = "请用中文详细描述一下图中的内容，比如时间，地点，人物，事情，人物心情等，如果有数据请提取出数据。" if self.lang.lower() == "chinese" else \
"Please describe the content of this picture, like where, when, who, what happen. If it has number data, please extract them out."
b64 = self.image2base64(image)
img = open(BytesIO(base64.b64decode(b64)))
input = [prompt,img]
res = self.model.generate_content(
input,
generation_config=gen_config,
)
return res.text,res.usage_metadata.total_token_count
class OpenRouterCV(Base):
def __init__(
self,
key,
model_name,
lang="Chinese",
base_url="https://openrouter.ai/api/v1/chat/completions",
):
self.model_name = model_name
self.lang = lang
self.base_url = "https://openrouter.ai/api/v1/chat/completions"
self.key = key
def describe(self, image, max_tokens=300):
b64 = self.image2base64(image)
response = requests.post(
url=self.base_url,
headers={
"Authorization": f"Bearer {self.key}",
},
data=json.dumps(
{
"model": self.model_name,
"messages": self.prompt(b64),
"max_tokens": max_tokens,
}
),
)
response = response.json()
return (
response["choices"][0]["message"]["content"].strip(),
response["usage"]["total_tokens"],
)
def prompt(self, b64):
return [
{
"role": "user",
"content": [
{
"type": "image_url",
"image_url": {"url": f"data:image/jpeg;base64,{b64}"},
},
{
"type": "text",
"text": (
"请用中文详细描述一下图中的内容，比如时间，地点，人物，事情，人物心情等，如果有数据请提取出数据。"
if self.lang.lower() == "chinese"
else "Please describe the content of this picture, like where, when, who, what happen. If it has number data, please extract them out."
),
},
],
}
]
class LocalCV(Base):
def __init__(self, key, model_name="glm-4v", lang="Chinese", **kwargs):
pass
def describe(self, image, max_tokens=1024):
return "", 0
|