upd

flash_vstream/constants.py

@@ -13,3 +13,6 @@ DEFAULT_IMAGE_PATCH_TOKEN = "<im_patch>"
 DEFAULT_IM_START_TOKEN = "<im_start>"
 DEFAULT_IM_END_TOKEN = "<im_end>"
 IMAGE_PLACEHOLDER = "<image-placeholder>"
+
+# Data Constants
+MAX_IMAGE_LENGTH = 3600

flash_vstream/serve/demo.py

@@ -3,7 +3,7 @@ from ..constants import *
 from ..conversation import conv_templates, SeparatorStyle
 from ..model.builder import load_pretrained_model
 from ..utils import disable_torch_init
-from ..mm_utils import tokenizer_image_token, KeywordsStoppingCriteria
+from ..mm_utils import tokenizer_image_token, KeywordsStoppingCriteria, get_model_name_from_path
 from PIL import Image
 import os
 from decord import VideoReader, cpu
@@ -13,7 +13,8 @@ import numpy as np
 class Chat:
     def __init__(self, model_path, conv_mode="simple", load_8bit=False, load_4bit=False):
         disable_torch_init()
-        self.tokenizer, self.model, self.image_processor, context_len = load_pretrained_model(model_path, None, model_name="ChatUniVi", load_8bit=load_8bit, load_4bit=load_4bit)
+        model_name = get_model_name_from_path(model_path)
+        self.tokenizer, self.model, self.image_processor, context_len = load_pretrained_model(model_path, None, model_name, load_8bit=load_8bit, load_4bit=load_4bit)
 
         mm_use_im_start_end = getattr(self.model.config, "mm_use_im_start_end", False)
         mm_use_im_patch_token = getattr(self.model.config, "mm_use_im_patch_token", True)
@@ -120,19 +121,16 @@ class Chat:
 
 title_markdown = ("""
 <div style="display: flex; justify-content: center; align-items: center; text-align: center;">
-  <a href="https://github.com/PKU-YuanGroup/Chat-UniVi" style="margin-right: 20px; text-decoration: none; display: flex; align-items: center;">
-    <img src="https://z1.ax1x.com/2023/11/22/pidlXh4.jpg" alt="Chat-UniVi🚀" style="max-width: 120px; height: auto;">
-  </a>
   <div>
-    <h1 >Chat-UniVi: Unified Visual Representation Empowers Large Language Models with Image and Video Understanding</h1>
+    <h1 >Flash-VStream: Memory-Based Real-Time Understanding for Long Video Streams</h1>
     <h5 style="margin: 0;">If you like our project, please give us a star ✨ on Github for the latest update.</h5>
   </div>
 </div>
 <div align="center">
     <div style="display:flex; gap: 0.25rem;" align="center">
-        <a href='https://github.com/PKU-YuanGroup/Chat-UniVi'><img src='https://img.shields.io/badge/Github-Code-blue'></a>
-        <a href="https://arxiv.org/pdf/2311.08046.pdf"><img src="https://img.shields.io/badge/Arxiv-2311.08046-red"></a>
-        <a href='https://github.com/PKU-YuanGroup/Chat-UniVi/stargazers'><img src='https://img.shields.io/github/stars/PKU-YuanGroup/Chat-UniVi.svg?style=social'></a>
+        <a href="https://invinciblewyq.github.io/vstream-page/"><img src='https://img.shields.io/badge/Project-Page-Green'></a>
+        <a href="https://arxiv.org/abs/2406.08085v1"><img src='https://img.shields.io/badge/Paper-Arxiv-red'></a>
+        <a href='https://github.com/IVGSZ/Flash-VStream'><img src='https://img.shields.io/badge/Github-Code-blue'></a>
     </div>
 </div>
 """)