Spaces:
Sleeping
Sleeping
MinxuanQin
commited on
Commit
•
de05d04
1
Parent(s):
f1c6918
add application file
Browse files
app.py
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import numpy as np
|
2 |
+
from PIL import Image
|
3 |
+
from transformers import ViltConfig, ViltProcessor, ViltForQuestionAnswering
|
4 |
+
|
5 |
+
import streamlit as st
|
6 |
+
|
7 |
+
st.title("Live demo of multimodal vqa")
|
8 |
+
|
9 |
+
config = ViltConfig.from_pretrained("dandelin/vilt-b32-finetuned-vqa")
|
10 |
+
|
11 |
+
processor = ViltProcessor.from_pretrained("dandelin/vilt-b32-finetuned-vqa")
|
12 |
+
model = ViltForQuestionAnswering.from_pretrained("Minqin/carets_vqa_finetuned")
|
13 |
+
|
14 |
+
uploaded_file = st.file_uploader("Please upload one image (jpg)", type="jpg")
|
15 |
+
|
16 |
+
question = st.text_input("Type here the question")
|
17 |
+
if uploaded_file is not None:
|
18 |
+
file_bytes = np.asarray(bytearray(uploaded_file.read()), dtype=np.unit8)
|
19 |
+
img = Image.fromarray(file_bytes)
|
20 |
+
|
21 |
+
encoding = processor(images=file_bytes, text=question, return_tensors="pt")
|
22 |
+
|
23 |
+
outputs = model(**encoding)
|
24 |
+
logits = outputs.logits
|
25 |
+
idx = logits.argmax(-1).item()
|
26 |
+
pred = model.config.id2label[idx]
|
27 |
+
|
28 |
+
st.text(f"Answer: {pred}")
|