[Python] Permission Denied Issue in Gradio

I am trying to create a gradio app which was working before now i have added the image functionality which analyzes the image and for text analysis the app was working fine but now it gives permission error:

import gradio as gr
from image import ImageAnalyzer
from cleaning import cleanText
from translate import translate_to_english
from sentiment import SentimentClassifier
from emotion import EmotionClassifier
from toxic import ToxicityClassifier
from racist import RacistClassifier

def classify_confidence(score):
if score >= 75:
return "High"
elif score >= 50:
return "Medium"
else:
return "Low"

def analyze_text(user_input):
try:
sentimentAnalyzer = SentimentClassifier()
emotionAnalyzer = EmotionClassifier()
toxicityAnalyzer = ToxicityClassifier()
racismAnalyzer = RacistClassifier()

cleanedText = cleanText(user_input)
translatedText = translate_to_english(cleanedText)

sentimentResult = sentimentAnalyzer.predict_sentiment(translatedText)
emotionResult = emotionAnalyzer.classifyEmotion(translatedText)
toxicResult = toxicityAnalyzer.predict(translatedText)
racismResult = racismAnalyzer.predict(translatedText)

# Processing sentiment results
sentiment_data = "\n".join(
[f"{label}: {data['confidence']}%" for label, data in sentimentResult.items()]
)

# Getting top 3 emotions without confidence classification
sorted_emotions = sorted(emotionResult.items(), key=lambda x: x[1], reverse=True)
top3_emotions = "\n".join(
[f"{emotion}: {score:.2f}%" for emotion, score in sorted_emotions[:3]]
)

# Processing toxicity and racism analysis with confidence classification
toxicity_analysis = "\n".join(
[f"{label}: {score:.2f}% ({classify_confidence(score)})" for label, score in toxicResult.items()]
)
racism_analysis = "\n".join(
[f"{label}: {score:.2f}% ({classify_confidence(score)})" for label, score in racismResult.items()]
)

return cleanedText, translatedText, sentiment_data, top3_emotions, toxicity_analysis, racism_analysis
except Exception as e:
return "", "", "", "", "", f"Unexpected error: {e}"

def analyze_image(image):
try:
analyzer = ImageAnalyzer()
return analyzer.def_image(image)
except Exception as e:
return "", "", f"Unexpected error: {e}"

def process_inputs(text, image):
if text and not image:
return analyze_text(text) + ("", "", "") # Empty placeholders for image outputs
elif image and not text:
return ("", "", "", "", "", "") + analyze_image(image) # Empty placeholders for text outputs
elif text and image:
return analyze_text(text) + analyze_image(image)
else:
return ("", "", "", "", "", "", "", "", "") # Empty results for both text and image inputs

def main():
interface = gr.Interface(
fn=process_inputs,
inputs=[gr.Textbox(lines=3, placeholder="Enter text to analyze..."), gr.Image(type="pil")],
outputs=[
gr.Textbox(label="Cleaned Text"),
gr.Textbox(label="Translated Text"),
gr.Textbox(label="Sentiment Analysis", lines=3),
gr.Textbox(label="Top 3 Emotion Detection", lines=3),
gr.Textbox(label="Toxicity Analysis", lines=3),
gr.Textbox(label="Racism Analysis", lines=3),
gr.Textbox(label="Image Analysis Results"),
gr.Image(label="ELA Image"),
gr.Image(label="Highlighted Forgery Image")
],
title="Text and Image Analysis Demo",
description="Analyze text for sentiment, emotions, toxicity, racism, and detect image forgery using AI models.",
)

interface.launch()

if __name__ == "__main__":
main()


and this is my code of ImageAnalyzer that i added:

import cv2
import numpy as np
from PIL import Image, ImageChops, ImageEnhance, UnidentifiedImageError
import os
import tensorflow as tf
import binascii

class ImageAnalyzer:
def __init__(self, model_path="../scripts/results/image/Model-2/MobileNetV2.keras"):
self.model = tf.keras.models.load_model(model_path)

def generate_unique_filename(self, extension=".jpg"):
return binascii.hexlify(os.urandom(8)).decode() + extension

def perform_ela(self, image_path, quality=90, threshold=30, save_path="output/"):
if not os.path.exists(save_path):
os.makedirs(save_path)

unique_filename = self.generate_unique_filename(".jpg")
temp_path = os.path.join(save_path, unique_filename)

original = Image.open(image_path).convert("RGB")
original.save(temp_path, "JPEG", quality=quality)
compressed = Image.open(temp_path)

ela_image = ImageChops.difference(original, compressed)
extrema = ela_image.getextrema()
max_diff = max([ex[1] for ex in extrema])

if max_diff == 0:
max_diff = 1

scale = 255.0 / max_diff
ela_image = ImageEnhance.Brightness(ela_image).enhance(scale)
ela_image_path = os.path.join(save_path, self.generate_unique_filename(".png"))
ela_image.save(ela_image_path)

ela_cv = cv2.imread(ela_image_path)
gray = cv2.cvtColor(ela_cv, cv2.COLOR_BGR2GRAY)
_, thresh = cv2.threshold(gray, threshold, 255, cv2.THRESH_BINARY)

total_pixels = thresh.size
forged_pixels = np.count_nonzero(thresh)
forged_percentage = (forged_pixels / total_pixels) * 100

highlight_path = os.path.join(save_path, self.generate_unique_filename(".png"))
cv2.imwrite(highlight_path, thresh)

is_forged = "Yes" if forged_percentage > 0.5 else "No"

return {
"forged": is_forged,
"percentage": round(forged_percentage, 2),
"ela_image": ela_image_path,
"highlighted_image": highlight_path
}

def predict_image(self, image_path):
image = Image.open(image_path).resize((224, 224))
image = np.array(image) / 255.0
image = np.expand_dims(image, axis=0)

prediction = self.model.predict(image)[0]
confidence_real = prediction[0]
confidence_fake = prediction[1]

label = "Real" if confidence_real > confidence_fake else "Fake"
confidence = max(confidence_real, confidence_fake) * 100

return {"label": label, "confidence": round(confidence, 2)}

def def_image(self, image):
try:
if not isinstance(image, Image.Image):
raise ValueError("Invalid image format. Please upload a valid image file.")

# Convert to RGB to remove alpha channels and ensure compatibility
image = image.convert("RGB")

# Save image as a temporary JPEG file
image_path = "image/temp_image.jpg"
image.save(image_path, format="JPEG")

# Perform ELA analysis and prediction
ela_result = self.perform_ela(image_path)
prediction_result = self.predict_image(image_path)

return {
"forged": ela_result["forged"],
"percentage": ela_result["percentage"],
"ela_image": ela_result["ela_image"],
"highlighted_image": ela_result["highlighted_image"],
"prediction_label": prediction_result["label"],
"prediction_confidence": prediction_result["confidence"]
}

except UnidentifiedImageError:
raise ValueError("Corrupt or unsupported image file. Please upload a valid image.")
except Exception as e:
raise ValueError(f"An unexpected error occurred: {e}")



Error:

PermissionError: [Errno 13] Permission denied: 'D:\\Uni\\FYP\\Model\\model\\Demo_Gradio'

I have tried creating a new directory and also made changes in the code but nothing really worked so far.

Continue reading...