Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -6,15 +6,20 @@ import time
|
|
6 |
from PIL import Image
|
7 |
import torch
|
8 |
import cv2
|
9 |
-
import spaces
|
10 |
-
|
|
|
11 |
|
12 |
processor = LlavaProcessor.from_pretrained(model_id)
|
13 |
|
14 |
model = LlavaForConditionalGeneration.from_pretrained(model_id, torch_dtype=torch.float16)
|
15 |
model.to("cuda")
|
16 |
|
17 |
-
|
|
|
|
|
|
|
|
|
18 |
video = cv2.VideoCapture(video_file)
|
19 |
total_frames = int(video.get(cv2.CAP_PROP_FRAME_COUNT))
|
20 |
interval = total_frames // num_frames
|
@@ -31,9 +36,16 @@ def sample_frames(video_file, num_frames) :
|
|
31 |
|
32 |
@spaces.GPU
|
33 |
def bot_streaming(message, history):
|
34 |
-
|
35 |
-
|
36 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
37 |
else:
|
38 |
# if there's no image uploaded for this turn, look for images in the past turns
|
39 |
# kept inside tuples, take the last one
|
@@ -41,28 +53,44 @@ def bot_streaming(message, history):
|
|
41 |
if type(hist[0])==tuple:
|
42 |
image = hist[0][0]
|
43 |
|
44 |
-
|
45 |
-
img = message["files"]
|
46 |
-
ext_buffer =f"'user\ntext': '{txt}', 'files': '{img}' assistant"
|
47 |
-
|
48 |
-
if image is None:
|
49 |
gr.Error("You need to upload an image or video for LLaVA to work.")
|
50 |
|
51 |
video_extensions = ("avi", "mp4", "mov", "mkv", "flv", "wmv", "mjpeg")
|
52 |
image_extensions = Image.registered_extensions()
|
53 |
image_extensions = tuple([ex for ex, f in image_extensions.items()])
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
59 |
|
60 |
-
|
61 |
-
|
62 |
-
|
|
|
|
|
63 |
|
64 |
inputs = processor(prompt, image, return_tensors="pt").to("cuda", torch.float16)
|
65 |
-
streamer = TextIteratorStreamer(processor, **{"skip_special_tokens": True})
|
66 |
generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=100)
|
67 |
generated_text = ""
|
68 |
|
@@ -75,15 +103,19 @@ def bot_streaming(message, history):
|
|
75 |
for new_text in streamer:
|
76 |
|
77 |
buffer += new_text
|
78 |
-
|
79 |
generated_text_without_prompt = buffer[len(ext_buffer):]
|
80 |
time.sleep(0.01)
|
81 |
yield generated_text_without_prompt
|
82 |
|
83 |
|
84 |
-
demo = gr.ChatInterface(fn=bot_streaming, title="LLaVA Interleave", examples=[
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
|
|
|
|
|
|
|
|
89 |
demo.launch(debug=True)
|
|
|
6 |
from PIL import Image
|
7 |
import torch
|
8 |
import cv2
|
9 |
+
import spaces
|
10 |
+
|
11 |
+
model_id = "llava-hf/llava-interleave-qwen-0.5b-hf"
|
12 |
|
13 |
processor = LlavaProcessor.from_pretrained(model_id)
|
14 |
|
15 |
model = LlavaForConditionalGeneration.from_pretrained(model_id, torch_dtype=torch.float16)
|
16 |
model.to("cuda")
|
17 |
|
18 |
+
|
19 |
+
def replace_video_with_images(text, frames):
|
20 |
+
return text.replace("<video>", "<image>" * frames)
|
21 |
+
|
22 |
+
def sample_frames(video_file, num_frames):
|
23 |
video = cv2.VideoCapture(video_file)
|
24 |
total_frames = int(video.get(cv2.CAP_PROP_FRAME_COUNT))
|
25 |
interval = total_frames // num_frames
|
|
|
36 |
|
37 |
@spaces.GPU
|
38 |
def bot_streaming(message, history):
|
39 |
+
|
40 |
+
txt = message.text
|
41 |
+
ext_buffer = f"user\n{txt} assistant"
|
42 |
+
|
43 |
+
if message.files:
|
44 |
+
if len(message.files) == 1:
|
45 |
+
image = [message.files[0].path]
|
46 |
+
# interleaved images or video
|
47 |
+
elif len(message.files) > 1:
|
48 |
+
image = [msg.path for msg in message.files]
|
49 |
else:
|
50 |
# if there's no image uploaded for this turn, look for images in the past turns
|
51 |
# kept inside tuples, take the last one
|
|
|
53 |
if type(hist[0])==tuple:
|
54 |
image = hist[0][0]
|
55 |
|
56 |
+
if message.files is None:
|
|
|
|
|
|
|
|
|
57 |
gr.Error("You need to upload an image or video for LLaVA to work.")
|
58 |
|
59 |
video_extensions = ("avi", "mp4", "mov", "mkv", "flv", "wmv", "mjpeg")
|
60 |
image_extensions = Image.registered_extensions()
|
61 |
image_extensions = tuple([ex for ex, f in image_extensions.items()])
|
62 |
+
if len(image) == 1:
|
63 |
+
if image[0].endswith(video_extensions):
|
64 |
+
|
65 |
+
image = sample_frames(image[0], 12)
|
66 |
+
image_tokens = "<image>" * 13
|
67 |
+
prompt = f"<|im_start|>user {image_tokens}\n{message.text}<|im_end|><|im_start|>assistant"
|
68 |
+
elif image[0].endswith(image_extensions):
|
69 |
+
image = Image.open(image[0]).convert("RGB")
|
70 |
+
prompt = f"<|im_start|>user <image>\n{message.text}<|im_end|><|im_start|>assistant"
|
71 |
+
|
72 |
+
elif len(image) > 1:
|
73 |
+
image_list = []
|
74 |
+
user_prompt = message.text
|
75 |
+
|
76 |
+
for img in image:
|
77 |
+
if img.endswith(image_extensions):
|
78 |
+
img = Image.open(img).convert("RGB")
|
79 |
+
image_list.append(img)
|
80 |
+
|
81 |
+
elif img.endswith(video_extensions):
|
82 |
+
frames = sample_frames(img, 6)
|
83 |
+
for frame in frames:
|
84 |
+
image_list.append(frame)
|
85 |
|
86 |
+
toks = "<image>" * len(image_list)
|
87 |
+
prompt = "<|im_start|>user"+ toks + f"\n{user_prompt}<|im_end|><|im_start|>assistant"
|
88 |
+
|
89 |
+
image = image_list
|
90 |
+
|
91 |
|
92 |
inputs = processor(prompt, image, return_tensors="pt").to("cuda", torch.float16)
|
93 |
+
streamer = TextIteratorStreamer(processor, **{"max_new_tokens": 200, "skip_special_tokens": True})
|
94 |
generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=100)
|
95 |
generated_text = ""
|
96 |
|
|
|
103 |
for new_text in streamer:
|
104 |
|
105 |
buffer += new_text
|
106 |
+
|
107 |
generated_text_without_prompt = buffer[len(ext_buffer):]
|
108 |
time.sleep(0.01)
|
109 |
yield generated_text_without_prompt
|
110 |
|
111 |
|
112 |
+
demo = gr.ChatInterface(fn=bot_streaming, title="LLaVA Interleave", examples=[
|
113 |
+
{"text": "What are these cats doing?", "files":["./cats.mp4"]},
|
114 |
+
{"text": "The input contains two videos, are the cats in this video and this video doing the same thing?", "files":["./cats_1.mp4", "./cats_2.mp4"]},
|
115 |
+
{"text": "What is on the flower?", "files":["./bee.jpg"]},
|
116 |
+
{"text": "There are two images in the input. What is the relationship between this image and this image?", "files":["./bee.jpg", "./depth-bee.png"]},
|
117 |
+
{"text": "How to make this pastry?", "files":["./baklava.png"]}],
|
118 |
+
textbox=gr.MultimodalTextbox(file_count="multiple"),
|
119 |
+
description="Try [LLaVA Interleave](https://huggingface.co/docs/transformers/main/en/model_doc/llava) in this demo (more specifically, the [Qwen-1.5-7B variant](https://huggingface.co/llava-hf/llava-interleave-qwen-7b-hf)). Upload an image or a video, and start chatting about it, or simply try one of the examples below. If you don't upload an image, you will receive an error. ",
|
120 |
+
stop_btn="Stop Generation", multimodal=True)
|
121 |
demo.launch(debug=True)
|