Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -2,57 +2,96 @@ import streamlit as st
|
|
2 |
from transformers import AutoProcessor, Qwen2VLForConditionalGeneration
|
3 |
from PIL import Image
|
4 |
import torch
|
|
|
|
|
5 |
|
6 |
# Load the processor and model directly
|
7 |
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
|
8 |
model = Qwen2VLForConditionalGeneration.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
|
9 |
|
|
|
|
|
|
|
|
|
10 |
# Streamlit app
|
11 |
-
st.title("
|
12 |
|
13 |
-
uploaded_file = st.file_uploader("Choose an image...", type=["jpg", "jpeg", "png"])
|
14 |
|
15 |
if uploaded_file is not None:
|
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 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
from transformers import AutoProcessor, Qwen2VLForConditionalGeneration
|
3 |
from PIL import Image
|
4 |
import torch
|
5 |
+
import cv2
|
6 |
+
import tempfile
|
7 |
|
8 |
# Load the processor and model directly
|
9 |
processor = AutoProcessor.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
|
10 |
model = Qwen2VLForConditionalGeneration.from_pretrained("Qwen/Qwen2-VL-2B-Instruct")
|
11 |
|
12 |
+
# Check if CUDA is available and set the device accordingly
|
13 |
+
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
|
14 |
+
model.to(device)
|
15 |
+
|
16 |
# Streamlit app
|
17 |
+
st.title("Media Description Generator")
|
18 |
|
19 |
+
uploaded_file = st.file_uploader("Choose an image or video...", type=["jpg", "jpeg", "png", "mp4", "avi", "mov"])
|
20 |
|
21 |
if uploaded_file is not None:
|
22 |
+
file_type = uploaded_file.type.split('/')[0]
|
23 |
+
|
24 |
+
if file_type == 'image':
|
25 |
+
# Open the image
|
26 |
+
image = Image.open(uploaded_file)
|
27 |
+
st.image(image, caption='Uploaded Image.', use_column_width=True)
|
28 |
+
st.write("Generating description...")
|
29 |
+
|
30 |
+
elif file_type == 'video':
|
31 |
+
# Save the uploaded video to a temporary file
|
32 |
+
tfile = tempfile.NamedTemporaryFile(delete=False)
|
33 |
+
tfile.write(uploaded_file.read())
|
34 |
+
|
35 |
+
# Open the video file
|
36 |
+
cap = cv2.VideoCapture(tfile.name)
|
37 |
+
|
38 |
+
# Extract the first frame
|
39 |
+
ret, frame = cap.read()
|
40 |
+
if not ret:
|
41 |
+
st.error("Failed to read the video file.")
|
42 |
+
st.stop()
|
43 |
+
else:
|
44 |
+
# Convert the frame to an image
|
45 |
+
image = Image.fromarray(cv2.cvtColor(frame, cv2.COLOR_BGR2RGB))
|
46 |
+
st.image(image, caption='First Frame of Uploaded Video.', use_column_width=True)
|
47 |
+
st.write("Generating description...")
|
48 |
+
|
49 |
+
# Release the video capture object
|
50 |
+
cap.release()
|
51 |
+
|
52 |
+
else:
|
53 |
+
st.error("Unsupported file type.")
|
54 |
+
st.stop()
|
55 |
+
|
56 |
+
# Add a text input for the user to ask a question
|
57 |
+
user_question = st.text_input("Ask a question about the image or video:")
|
58 |
+
|
59 |
+
if user_question:
|
60 |
+
messages = [
|
61 |
+
{
|
62 |
+
"role": "user",
|
63 |
+
"content": [
|
64 |
+
{
|
65 |
+
"type": "image",
|
66 |
+
"image": image,
|
67 |
+
},
|
68 |
+
{"type": "text", "text": user_question},
|
69 |
+
],
|
70 |
+
}
|
71 |
+
]
|
72 |
+
|
73 |
+
# Preparation for inference
|
74 |
+
text = processor.apply_chat_template(
|
75 |
+
messages, tokenize=False, add_generation_prompt=True
|
76 |
+
)
|
77 |
+
|
78 |
+
# Pass the image to the processor
|
79 |
+
inputs = processor(
|
80 |
+
text=[text],
|
81 |
+
images=[image],
|
82 |
+
padding=True,
|
83 |
+
return_tensors="pt",
|
84 |
+
)
|
85 |
+
inputs = inputs.to(device) # Ensure inputs are on the same device as the model
|
86 |
+
|
87 |
+
# Inference: Generation of the output
|
88 |
+
generated_ids = model.generate(**inputs, max_new_tokens=128)
|
89 |
+
generated_ids_trimmed = [
|
90 |
+
out_ids[len(in_ids) :] for in_ids, out_ids in zip(inputs.input_ids, generated_ids)
|
91 |
+
]
|
92 |
+
output_text = processor.batch_decode(
|
93 |
+
generated_ids_trimmed, skip_special_tokens=True, clean_up_tokenization_spaces=False
|
94 |
+
)
|
95 |
+
|
96 |
+
st.write("Description:")
|
97 |
+
st.write(output_text[0])
|