Spaces:
Paused
Paused
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,8 +2,6 @@ import gradio as gr
|
|
| 2 |
from PIL import Image
|
| 3 |
import torch
|
| 4 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
| 5 |
-
import magic
|
| 6 |
-
import mimetypes
|
| 7 |
import cv2
|
| 8 |
import numpy as np
|
| 9 |
import io
|
|
@@ -22,16 +20,6 @@ model = AutoModelForCausalLM.from_pretrained("ManishThota/SparrowVQE",
|
|
| 22 |
tokenizer = AutoTokenizer.from_pretrained("ManishThota/SparrowVQE", trust_remote_code=True)
|
| 23 |
|
| 24 |
|
| 25 |
-
def get_file_type_from_bytes(file_bytes):
|
| 26 |
-
"""Determine whether a file is an image or a video based on its MIME type from bytes."""
|
| 27 |
-
mime = magic.Magic(mime=True)
|
| 28 |
-
mimetype = mime.from_buffer(file_bytes)
|
| 29 |
-
if mimetype.startswith('image'):
|
| 30 |
-
return 'image'
|
| 31 |
-
elif mimetype.startswith('video'):
|
| 32 |
-
return 'video'
|
| 33 |
-
return 'unknown'
|
| 34 |
-
|
| 35 |
def process_video(video_bytes):
|
| 36 |
"""Extracts frames from the video, 1 per second."""
|
| 37 |
video = cv2.VideoCapture(io.BytesIO(video_bytes))
|
|
@@ -46,15 +34,12 @@ def process_video(video_bytes):
|
|
| 46 |
return frames[:4] # Return the first 4 frames
|
| 47 |
|
| 48 |
|
| 49 |
-
def predict_answer(
|
| 50 |
-
|
| 51 |
-
file_type = get_file_type_from_bytes(file)
|
| 52 |
|
| 53 |
-
if
|
| 54 |
# Process as an image
|
| 55 |
-
image =
|
| 56 |
-
|
| 57 |
-
input_ids = tokenizer(text, return_tensors='pt').input_ids.to(device)
|
| 58 |
image_tensor = model.image_preprocess(frame)
|
| 59 |
|
| 60 |
#Generate the answer
|
|
@@ -66,13 +51,13 @@ def predict_answer(file, question, max_tokens=100):
|
|
| 66 |
|
| 67 |
return tokenizer.decode(output_ids[input_ids.shape[1]:], skip_special_tokens=True).strip()
|
| 68 |
|
| 69 |
-
elif
|
| 70 |
# Process as a video
|
| 71 |
-
frames = process_video(
|
| 72 |
answers = []
|
| 73 |
for frame in frames:
|
| 74 |
frame = Image.open(frame).convert("RGB")
|
| 75 |
-
input_ids = tokenizer(
|
| 76 |
image_tensor = model.image_preprocess(frame)
|
| 77 |
|
| 78 |
# Generate the answer
|
|
@@ -90,45 +75,19 @@ def predict_answer(file, question, max_tokens=100):
|
|
| 90 |
return "Unsupported file type. Please upload an image or video."
|
| 91 |
|
| 92 |
|
| 93 |
-
|
| 94 |
-
|
| 95 |
-
|
| 96 |
-
# def predict_answer(image, question, max_tokens=100):
|
| 97 |
-
# #Set inputs
|
| 98 |
-
# text = f"A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: <image>\n{question}? ASSISTANT:"
|
| 99 |
-
# image = image.convert("RGB")
|
| 100 |
-
|
| 101 |
-
# input_ids = tokenizer(text, return_tensors='pt').input_ids.to(device)
|
| 102 |
-
# image_tensor = model.image_preprocess(image)
|
| 103 |
-
|
| 104 |
-
# #Generate the answer
|
| 105 |
-
# output_ids = model.generate(
|
| 106 |
-
# input_ids,
|
| 107 |
-
# max_new_tokens=max_tokens,
|
| 108 |
-
# images=image_tensor,
|
| 109 |
-
# use_cache=True)[0]
|
| 110 |
-
|
| 111 |
-
# return tokenizer.decode(output_ids[input_ids.shape[1]:], skip_special_tokens=True).strip()
|
| 112 |
|
| 113 |
-
|
| 114 |
-
|
|
|
|
| 115 |
return answer
|
| 116 |
|
| 117 |
-
|
| 118 |
-
# examples = [["data/week_01_page_024.png", 'Can you explain the slide?',100],
|
| 119 |
-
# ["data/week_03_page_091.png", 'Can you explain the slide?',100],
|
| 120 |
-
# ["data/week_01_page_062.png", 'Are the training images labeled?',100],
|
| 121 |
-
# ["data/week_05_page_027.png", 'What is meant by eigenvalue multiplicity?',100],
|
| 122 |
-
# ["data/week_05_page_030.png", 'What does K represent?',100],
|
| 123 |
-
# ["data/week_15_page_046.png", 'How are individual heterogeneous models trained?',100],
|
| 124 |
-
# ["data/week_15_page_021.png", 'How does Bagging affect error?',100],
|
| 125 |
-
# ["data/week_15_page_037.png", "What does the '+' and '-' represent?",100]]
|
| 126 |
|
| 127 |
# Define the Gradio interface
|
| 128 |
iface = gr.Interface(
|
| 129 |
fn=gradio_predict,
|
| 130 |
-
inputs=[gr.
|
| 131 |
-
|
| 132 |
gr.Textbox(label="Question", placeholder="e.g. Can you explain the slide?", scale=4),
|
| 133 |
gr.Slider(2, 500, value=25, label="Token Count", info="Choose between 2 and 500")],
|
| 134 |
outputs=gr.TextArea(label="Answer"),
|
|
|
|
| 2 |
from PIL import Image
|
| 3 |
import torch
|
| 4 |
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
|
|
|
| 5 |
import cv2
|
| 6 |
import numpy as np
|
| 7 |
import io
|
|
|
|
| 20 |
tokenizer = AutoTokenizer.from_pretrained("ManishThota/SparrowVQE", trust_remote_code=True)
|
| 21 |
|
| 22 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 23 |
def process_video(video_bytes):
|
| 24 |
"""Extracts frames from the video, 1 per second."""
|
| 25 |
video = cv2.VideoCapture(io.BytesIO(video_bytes))
|
|
|
|
| 34 |
return frames[:4] # Return the first 4 frames
|
| 35 |
|
| 36 |
|
| 37 |
+
def predict_answer(image, video, question, max_tokens=100):
|
|
|
|
|
|
|
| 38 |
|
| 39 |
+
if image:
|
| 40 |
# Process as an image
|
| 41 |
+
image = image.convert("RGB")
|
| 42 |
+
input_ids = tokenizer(question, return_tensors='pt').input_ids.to(device)
|
|
|
|
| 43 |
image_tensor = model.image_preprocess(frame)
|
| 44 |
|
| 45 |
#Generate the answer
|
|
|
|
| 51 |
|
| 52 |
return tokenizer.decode(output_ids[input_ids.shape[1]:], skip_special_tokens=True).strip()
|
| 53 |
|
| 54 |
+
elif video:
|
| 55 |
# Process as a video
|
| 56 |
+
frames = process_video(video)
|
| 57 |
answers = []
|
| 58 |
for frame in frames:
|
| 59 |
frame = Image.open(frame).convert("RGB")
|
| 60 |
+
input_ids = tokenizer(question, return_tensors='pt').input_ids.to(device)
|
| 61 |
image_tensor = model.image_preprocess(frame)
|
| 62 |
|
| 63 |
# Generate the answer
|
|
|
|
| 75 |
return "Unsupported file type. Please upload an image or video."
|
| 76 |
|
| 77 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 78 |
|
| 79 |
+
|
| 80 |
+
def gradio_predict(image, video, question, max_tokens):
|
| 81 |
+
answer = predict_answer(image, video, question, max_tokens)
|
| 82 |
return answer
|
| 83 |
|
| 84 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 85 |
|
| 86 |
# Define the Gradio interface
|
| 87 |
iface = gr.Interface(
|
| 88 |
fn=gradio_predict,
|
| 89 |
+
inputs=[gr.Image(type="pil", label="Upload or Drag an Image"),
|
| 90 |
+
gr.Video(label="upload your video here"),
|
| 91 |
gr.Textbox(label="Question", placeholder="e.g. Can you explain the slide?", scale=4),
|
| 92 |
gr.Slider(2, 500, value=25, label="Token Count", info="Choose between 2 and 500")],
|
| 93 |
outputs=gr.TextArea(label="Answer"),
|