Commit
ยท
e6856f6
1
Parent(s):
94b8dff
app and caption generator file running and working
Browse files- app.py +32 -0
- assets/an_artist_would_call.jpeg +0 -0
- assets/drawing.jpeg +0 -0
- assets/generate.jpeg +0 -0
- assets/thinking.jpeg +0 -0
- assets/to be added text.txt +10 -0
- pages/Caption_Generator.py +120 -0
- requirements.txt +0 -0
app.py
ADDED
|
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import streamlit as st
|
| 2 |
+
|
| 3 |
+
def main():
|
| 4 |
+
|
| 5 |
+
st.markdown('<div style="display: flex; justify-content: center;"><p style="font-size: 30px; font-weight: bold;">Artwork Caption Generation</p></div>', unsafe_allow_html=True)
|
| 6 |
+
col_left, col_right = st.columns(2)
|
| 7 |
+
|
| 8 |
+
with col_left:
|
| 9 |
+
col11, col12 = st.columns(2)
|
| 10 |
+
with col11:
|
| 11 |
+
st.image(image="assets\drawing.jpeg")
|
| 12 |
+
with col12:
|
| 13 |
+
st.write("<div style='text-align:right;'>Witness the birth of creativity. Our website breathes life into your artistic visions.</div>", unsafe_allow_html=True)
|
| 14 |
+
col21, col22 = st.columns(2)
|
| 15 |
+
with col21:
|
| 16 |
+
st.write("<div style='text-align:left;'>Delve into the depths of inspiration. Explore our platform to give voice to your masterpiece.</div>", unsafe_allow_html=True)
|
| 17 |
+
with col22:
|
| 18 |
+
st.image(image="assets\\thinking.jpeg")
|
| 19 |
+
col31, col32 = st.columns(2)
|
| 20 |
+
with col31:
|
| 21 |
+
st.image(image="assets\generate.jpeg")
|
| 22 |
+
with col32:
|
| 23 |
+
st.write("<div style='text-align:right;'>Crafting the narrative of your creation. Unleash your imagination with our caption generator for artwork.</div>", unsafe_allow_html=True)
|
| 24 |
+
|
| 25 |
+
with col_right:
|
| 26 |
+
st.subheader('How to use')
|
| 27 |
+
|
| 28 |
+
if st.button(label="Generator", type="primary"):
|
| 29 |
+
st.switch_page(page='pages/Caption_Generator.py')
|
| 30 |
+
|
| 31 |
+
if __name__ == "__main__":
|
| 32 |
+
main()
|
assets/an_artist_would_call.jpeg
ADDED
|
assets/drawing.jpeg
ADDED
|
assets/generate.jpeg
ADDED
|
assets/thinking.jpeg
ADDED
|
assets/to be added text.txt
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
|
| 2 |
+
|
| 3 |
+
|
| 4 |
+
Witness the birth of creativity. Our website breathes life into your artistic visions.
|
| 5 |
+
|
| 6 |
+
|
| 7 |
+
Delve into the depths of inspiration. Explore our platform to give voice to your masterpiece.
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
Crafting the narrative of your creation. Unleash your imagination with our caption generator for artwork.
|
pages/Caption_Generator.py
ADDED
|
@@ -0,0 +1,120 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import torch
|
| 2 |
+
import numpy as np
|
| 3 |
+
from PIL import Image
|
| 4 |
+
import streamlit as st
|
| 5 |
+
from torchvision.transforms import v2
|
| 6 |
+
from transformers import GenerationConfig
|
| 7 |
+
from transformers import GPT2TokenizerFast
|
| 8 |
+
from transformers import ViTImageProcessor
|
| 9 |
+
from transformers import VisionEncoderDecoderModel
|
| 10 |
+
|
| 11 |
+
# Page configuration settings
|
| 12 |
+
st.set_page_config(
|
| 13 |
+
layout="centered",
|
| 14 |
+
page_title="Generate Caption",
|
| 15 |
+
initial_sidebar_state="auto",
|
| 16 |
+
)
|
| 17 |
+
|
| 18 |
+
# Initializing session state keys
|
| 19 |
+
if all(key not in st.session_state.keys() for key in ("generate", "image")):
|
| 20 |
+
st.session_state["generate"] = False
|
| 21 |
+
st.session_state["image"] = None
|
| 22 |
+
|
| 23 |
+
# Loading necessary resources and caching them
|
| 24 |
+
@st.cache_resource(show_spinner="Loading Resources...")
|
| 25 |
+
def loadResources():
|
| 26 |
+
encoder = 'microsoft/swin-base-patch4-window7-224-in22k'
|
| 27 |
+
decoder = 'gpt2'
|
| 28 |
+
|
| 29 |
+
model = VisionEncoderDecoderModel.from_encoder_decoder_pretrained(
|
| 30 |
+
encoder, decoder
|
| 31 |
+
)
|
| 32 |
+
|
| 33 |
+
processor = ViTImageProcessor.from_pretrained(encoder)
|
| 34 |
+
tokenizer = GPT2TokenizerFast.from_pretrained(decoder)
|
| 35 |
+
|
| 36 |
+
if 'gpt2' in decoder:
|
| 37 |
+
tokenizer.pad_token = tokenizer.eos_token
|
| 38 |
+
model.config.eos_token_id = tokenizer.eos_token_id
|
| 39 |
+
model.config.pad_token_id = tokenizer.pad_token_id
|
| 40 |
+
model.config.decoder_start_token_id = tokenizer.bos_token_id
|
| 41 |
+
else:
|
| 42 |
+
model.config.decoder_start_token_id = tokenizer.cls_token_id
|
| 43 |
+
model.config.pad_token_id = tokenizer.pad_token_id
|
| 44 |
+
|
| 45 |
+
model = torch.load("generator_model.pkl", map_location=torch.device("cpu"))
|
| 46 |
+
model.eval()
|
| 47 |
+
return processor, tokenizer, model
|
| 48 |
+
|
| 49 |
+
# Pre-processing image and caching
|
| 50 |
+
@st.cache_data
|
| 51 |
+
def preprocess_image(_processor, image):
|
| 52 |
+
transforms = v2.Compose([
|
| 53 |
+
v2.Resize(size=(224,224)),
|
| 54 |
+
v2.ToDtype(torch.float32, scale = True),
|
| 55 |
+
])
|
| 56 |
+
image = transforms(image)
|
| 57 |
+
img = _processor(image, return_tensors = 'pt')
|
| 58 |
+
return img
|
| 59 |
+
|
| 60 |
+
# Generating caption and caching
|
| 61 |
+
@st.cache_data
|
| 62 |
+
def get_caption(_processor, _tokenizer, _model, image):
|
| 63 |
+
image = preprocess_image(_processor, image)
|
| 64 |
+
output = _model.generate(
|
| 65 |
+
**image,
|
| 66 |
+
generation_config = GenerationConfig(
|
| 67 |
+
pad_token_id = _tokenizer.pad_token_id
|
| 68 |
+
)
|
| 69 |
+
)
|
| 70 |
+
|
| 71 |
+
caption = _tokenizer.batch_decode(
|
| 72 |
+
output,
|
| 73 |
+
skip_special_tokens = True
|
| 74 |
+
)
|
| 75 |
+
|
| 76 |
+
return caption[0]
|
| 77 |
+
|
| 78 |
+
# Displaying elements
|
| 79 |
+
def DisplayInteractionElements():
|
| 80 |
+
st.markdown('<div style="display: flex; justify-content: center;"><p style="font-size: 40px; font-weight: bold;">๐ Caption Generator ๐</p></div>', unsafe_allow_html=True)
|
| 81 |
+
st.file_uploader(accept_multiple_files=False, label='Upload an Image', type=['jpg', 'jpeg', 'png'], key="image_uploader")
|
| 82 |
+
|
| 83 |
+
if st.session_state['image_uploader']:
|
| 84 |
+
image = st.session_state['image_uploader']
|
| 85 |
+
im_file = Image.open(image).convert("RGB")
|
| 86 |
+
im_file = np.array(im_file)
|
| 87 |
+
|
| 88 |
+
st.session_state['image'] = im_file
|
| 89 |
+
|
| 90 |
+
col1, col2, col3 = st.columns(3)
|
| 91 |
+
|
| 92 |
+
col2.image(image=image, caption='Uploaded Image')
|
| 93 |
+
|
| 94 |
+
st.button(label='Generate Caption', use_container_width=True, type='primary', on_click=generateCaption)
|
| 95 |
+
|
| 96 |
+
# Triggering generate state
|
| 97 |
+
def generateCaption():
|
| 98 |
+
st.session_state['generate'] = True
|
| 99 |
+
|
| 100 |
+
def main():
|
| 101 |
+
|
| 102 |
+
DisplayInteractionElements()
|
| 103 |
+
|
| 104 |
+
processor, tokenizer, model = loadResources()
|
| 105 |
+
|
| 106 |
+
if not st.session_state['image_uploader']:
|
| 107 |
+
st.session_state['generate'] = False
|
| 108 |
+
|
| 109 |
+
if st.session_state['generate'] and st.session_state['image_uploader']:
|
| 110 |
+
caption = get_caption(processor, tokenizer, model, st.session_state['image'])
|
| 111 |
+
st.markdown(f'<div style="display: flex; justify-content: center;"><p style="font-size: 35px; font-weight: bold; color: blue;">{caption}</p></div>', unsafe_allow_html = True)
|
| 112 |
+
|
| 113 |
+
|
| 114 |
+
if __name__ == "__main__":
|
| 115 |
+
main()
|
| 116 |
+
|
| 117 |
+
|
| 118 |
+
|
| 119 |
+
|
| 120 |
+
|
requirements.txt
ADDED
|
Binary file (214 Bytes). View file
|
|
|