Update app.py

app.py

@@ -1,257 +1,93 @@
 import gradio as gr
 import torch
-from transformers import AutoFeatureExtractor, AutoModelForImageClassification, pipeline
-#from transformers import pipeline
-import os 
+from transformers import AutoFeatureExtractor, AutoModelForImageClassification
 from numpy import exp
-import pandas as  pd
+import pandas as pd
 from PIL import Image
 import urllib.request 
 import uuid
-uid=uuid.uuid4()
 
-models=[
+uid = uuid.uuid4()
+models = [
     "cmckinle/sdxl-flux-detector",
     "umm-maybe/AI-image-detector",
     "Organika/sdxl-detector",
-    #"arnolfokam/ai-generated-image-detector",
 ]
-
-pipe0 = pipeline("image-classification", f"{models[0]}")
-pipe1 = pipeline("image-classification", f"{models[1]}")
-pipe2 = pipeline("image-classification", f"{models[2]}")
- 
-
-fin_sum=[] 
-def image_classifier0(image):
-    labels = ["AI","Real"]
-    outputs = pipe0(image)
-    results = {}
-    result_test={}
-    for idx,result in enumerate(outputs):
-        results[labels[idx]] = outputs[idx]['score']
-    #print (result_test)
-    #for result in outputs:
-    #    results[result['label']] = result['score']    
-    #print (results) 
-    fin_sum.append(results)
-    return results
-def image_classifier1(image):
-    labels = ["AI","Real"]
-    outputs = pipe1(image)
-    results = {}
-    result_test={}
-    for idx,result in enumerate(outputs):
-        results[labels[idx]] = outputs[idx]['score']
-    #print (result_test)    
-    #for result in outputs:
-    #    results[result['label']] = result['score']
-    #print (results)
-    fin_sum.append(results)
-    return results
-def image_classifier2(image):
-    labels = ["AI","Real"]
-    outputs = pipe2(image)
-    results = {}
-    result_test={}
-    for idx,result in enumerate(outputs):
-        results[labels[idx]] = outputs[idx]['score']
-    #print (result_test)    
-    #for result in outputs:
-    #    results[result['label']] = result['score']
-    #print (results) 
-    fin_sum.append(results)
-    return results
+results_store = []
 
 def softmax(vector):
- e = exp(vector)
- return e / e.sum()
+    e = exp(vector)
+    return e / e.sum()
 
-     
-
-def aiornot0(image):    
-    labels = ["AI", "Real"]
-    mod=models[0]
-    feature_extractor0 = AutoFeatureExtractor.from_pretrained(mod)
-    model0 = AutoModelForImageClassification.from_pretrained(mod)
-    input = feature_extractor0(image, return_tensors="pt")
-    with torch.no_grad():
-        outputs = model0(**input)
-        logits = outputs.logits
-        probability = softmax(logits)
-        px = pd.DataFrame(probability.numpy())
-    prediction = logits.argmax(-1).item()
-    label = labels[prediction]
-    html_out = f"""
-    <h1>This image is likely: {label}</h1><br><h3>
- 
-    Probabilites:<br>
-    Real: {px[1][0]}<br>
-    AI: {px[0][0]}"""
-    results = {}
-    for idx,result in enumerate(px):
-        results[labels[idx]] = px[idx][0]
-    #results[labels['label']] = result['score']
-    fin_sum.append(results)
-    return gr.HTML.update(html_out),results
-def aiornot1(image):    
-    labels = ["AI", "Real"]
-    mod=models[1]
-    feature_extractor1 = AutoFeatureExtractor.from_pretrained(mod)
-    model1 = AutoModelForImageClassification.from_pretrained(mod)
-    input = feature_extractor1(image, return_tensors="pt")
-    with torch.no_grad():
-        outputs = model1(**input)
-        logits = outputs.logits
-        probability = softmax(logits)
-        px = pd.DataFrame(probability.numpy())
-    prediction = logits.argmax(-1).item()
-    label = labels[prediction]
-    html_out = f"""
-    <h1>This image is likely: {label}</h1><br><h3>
-  
-    Probabilites:<br>
-    Real: {px[1][0]}<br>
-    AI: {px[0][0]}"""
-    results = {}
-    for idx,result in enumerate(px):
-        results[labels[idx]] = px[idx][0]
-    #results[labels['label']] = result['score']
-    fin_sum.append(results)
-    return gr.HTML.update(html_out),results    
-def aiornot2(image):    
-    labels = ["Real", "AI"]
-    mod=models[2]
-    feature_extractor2 = AutoFeatureExtractor.from_pretrained(mod)
-    #feature_extractor2 = AutoFeatureExtractor.from_pretrained("microsoft/resnet-50")
-    model2 = AutoModelForImageClassification.from_pretrained(mod)
-    input = feature_extractor2(image, return_tensors="pt")
+def aiornot(image, model_index):    
+    mod = models[model_index]
+    feature_extractor = AutoFeatureExtractor.from_pretrained(mod)
+    model = AutoModelForImageClassification.from_pretrained(mod)
+    input = feature_extractor(image, return_tensors="pt")
     with torch.no_grad():
-        outputs = model2(**input)
+        outputs = model(**input)
         logits = outputs.logits
         probability = softmax(logits)
         px = pd.DataFrame(probability.numpy())
-    prediction = logits.argmax(-1).item()
-    label = labels[prediction]
+    
+    if model_index == 2:  # Organika model
+        real_prob, ai_prob = px[0][0], px[1][0]
+        label = "Real" if real_prob > ai_prob else "AI"
+    else:
+        ai_prob, real_prob = px[0][0], px[1][0]
+        label = "AI" if ai_prob > real_prob else "Real"
+    
     html_out = f"""
     <h1>This image is likely: {label}</h1><br><h3>
-  
-    Probabilites:<br>
-    Real: {px[0][0]}<br>
-    AI: {px[1][0]}"""
-
-    results = {}
-    for idx,result in enumerate(px):
-        results[labels[idx]] = px[idx][0]
-    #results[labels['label']] = result['score']
-    fin_sum.append(results)
-        
-    return gr.HTML.update(html_out),results
+    Probabilities:<br>
+    Real: {real_prob:.4f}<br>
+    AI: {ai_prob:.4f}"""
+    
+    results = {"Real": real_prob, "AI": ai_prob}
+    results_store.append(results)
+    return gr.HTML.update(html_out), results
 
 def load_url(url):
     try:
-        urllib.request.urlretrieve( 
-            f'{url}', 
-            f"{uid}tmp_im.png")         
+        urllib.request.urlretrieve(f'{url}', f"{uid}tmp_im.png")         
         image = Image.open(f"{uid}tmp_im.png")
         mes = "Image Loaded"
     except Exception as e:
-        image=None
-        mes=f"Image not Found<br>Error: {e}"
-    return image,mes
-
-def tot_prob():
-    try:
-        fin_out = fin_sum[0]["Real"]+fin_sum[1]["Real"]+fin_sum[2]["Real"]+fin_sum[3]["Real"]+fin_sum[4]["Real"]+fin_sum[5]["Real"]
-        fin_out = fin_out/6
-        fin_sub = 1-fin_out
-        out={
-            "Real":f"{fin_out}",
-            "AI":f"{fin_sub}"
-        }
-        #fin_sum.clear()
-        #print (fin_out)
-        return out
-    except Exception as e:
-        pass
-        print (e)
-        return None
-def fin_clear():
-    fin_sum.clear()
-    return None
-
-def upd(image):
-    print (image)
-    rand_im = uuid.uuid4()
-    image.save(f"{rand_im}-vid_tmp_proc.png")
-    out = Image.open(f"{rand_im}-vid_tmp_proc.png")
+        image = None
+        mes = f"Image not Found<br>Error: {e}"
+    return image, mes
 
-    #image.save(f"{rand_im}-vid_tmp_proc.png")
-    #out = os.path.abspath(f"{rand_im}-vid_tmp_proc.png")
-    #out_url = f'https://omnibus_AI_or_Not_dev.hf.space/file={out}'
-    #out_url = f"{rand_im}-vid_tmp_proc.png"
-    return out 
+def calculate_final_prob():
+    if not results_store:
+        return {"Real": "N/A", "AI": "N/A"}
+    fin_out = sum(result["Real"] for result in results_store) / len(results_store)
+    return {
+        "Real": f"{fin_out:.4f}",
+        "AI": f"{1 - fin_out:.4f}"
+    }
 
-    
 with gr.Blocks() as app:
-    gr.Markdown("""<center><h1>AI Image Detector<br><h4>(Test Demo - accuracy varies by model)""")
+    gr.Markdown("""<center><h1>AI Image Detector<br><h4>(Test Demo - accuracy varies by model)</h4></center>""")
     with gr.Column():
         inp = gr.Image(type='pil')
-        in_url=gr.Textbox(label="Image URL")
+        in_url = gr.Textbox(label="Image URL")
         with gr.Row():
-            load_btn=gr.Button("Load URL")
+            load_btn = gr.Button("Load URL")
             btn = gr.Button("Detect AI")
-        mes = gr.HTML("""""")
-    with gr.Group():    
-        with gr.Row():
-            fin=gr.Label(label="Final Probability")       
+        mes = gr.HTML()
+    with gr.Group():
         with gr.Row():
-            with gr.Box():
-                lab0 = gr.HTML(f"""<b>Testing on Model: <a href='https://huggingface.co/{models[0]}'>{models[0]}</a></b>""")
-                nun0 = gr.HTML("""""")
-            with gr.Box():
-                lab1 = gr.HTML(f"""<b>Testing on Model: <a href='https://huggingface.co/{models[1]}'>{models[1]}</a></b>""")
-                nun1 = gr.HTML("""""")
-            with gr.Box():
-                lab2 = gr.HTML(f"""<b>Testing on Model: <a href='https://huggingface.co/{models[2]}'>{models[2]}</a></b>""")
-                nun2 = gr.HTML("""""")
-                
+            fin = gr.Label(label="Final Probability")
         with gr.Row():
-            with gr.Box():
-                n_out0=gr.Label(label="Output")
-                outp0 = gr.HTML("""""")
-            with gr.Box():
-                n_out1=gr.Label(label="Output")
-                outp1 = gr.HTML("""""")
-            with gr.Box():
-                n_out2=gr.Label(label="Output")
-                outp2 = gr.HTML("""""")    
-        with gr.Row():
-            with gr.Box():
-                n_out3=gr.Label(label="Output")
-                outp3 = gr.HTML("""""")
-            with gr.Box():
-                n_out4=gr.Label(label="Output")
-                outp4 = gr.HTML("""""")
-            with gr.Box():
-                n_out5=gr.Label(label="Output")
-                outp5 = gr.HTML("""""")    
-    hid_box=gr.Textbox(visible=False)  
-    hid_im = gr.Image(type="pil",visible=False)
-    def echo(inp):
-        return inp
-
-    #inp.change(echo,inp,hid_im).then(upd,hid_im,inp)
-    
-    btn.click(fin_clear,None,fin,show_progress=False)
-    load_btn.click(load_url,in_url,[inp,mes])
-    
-    btn.click(aiornot0,[inp],[outp0,n_out0]).then(tot_prob,None,fin,show_progress=False)
-    btn.click(aiornot1,[inp],[outp1,n_out1]).then(tot_prob,None,fin,show_progress=False)
-    btn.click(aiornot2,[inp],[outp2,n_out2]).then(tot_prob,None,fin,show_progress=False)
-    
-    btn.click(image_classifier0,[inp],[n_out3]).then(tot_prob,None,fin,show_progress=False)
-    btn.click(image_classifier1,[inp],[n_out4]).then(tot_prob,None,fin,show_progress=False)
-    btn.click(image_classifier2,[inp],[n_out5]).then(tot_prob,None,fin,show_progress=False)
+            for i, model in enumerate(models):
+                with gr.Column():
+                    gr.HTML(f"""<b>Testing on Model: <a href='https://huggingface.co/{model}'>{model}</a></b>""")
+                    output_html = gr.HTML()
+                    output_label = gr.Label(label="Output")
+                    btn.click(aiornot, inputs=[inp, gr.Number(value=i, visible=False)], outputs=[output_html, output_label])
+                    if i == len(models) - 1:  # After the last model, calculate final probability
+                        btn.click(lambda: results_store.clear(), outputs=None)  # Clear results before new detection
+                        btn.click(calculate_final_prob, outputs=fin)
+    load_btn.click(load_url, in_url, [inp, mes])
 
-app.launch(show_api=False,max_threads=24)
+app.launch(show_api=False, max_threads=24)