update app

app.py

@@ -82,7 +82,6 @@ model_m = Qwen2_5_VLForConditionalGeneration.from_pretrained(
     torch_dtype=torch.float16
 ).to(device).eval()
 
-
 MODEL_MAP = {
     "Nanonets-OCR2-3B": (processor_v, model_v),
     "olmOCR-7B-0725": (processor_w, model_w),
@@ -90,6 +89,7 @@ MODEL_MAP = {
     "Aya-Vision-8B": (processor_a, model_a),
     "Qwen2-VL-OCR-2B": (processor_x, model_x),
 }
+
 MODEL_CHOICES = list(MODEL_MAP.keys())
 
 image_examples = [
@@ -97,7 +97,7 @@ image_examples = [
     {"query": "Run OCR on the image and ensure high accuracy.", "image": "examples/4.jpg", "model": "olmOCR-7B-0725"},
     {"query": "Conduct OCR on the image with exact text recognition.", "image": "examples/2.jpg", "model": "RolmOCR-7B"},
     {"query": "Perform precise OCR extraction on the image.", "image": "examples/1.jpg", "model": "Qwen2-VL-OCR-2B"},
-    {"query": "Describe the image and read visible text in detail.", "image": "examples/3.jpg", "model": "Aya-Vision-8B"},
+    {"query": "Describe the visual content and extract visible text from the image.", "image": "examples/3.jpg", "model": "Aya-Vision-8B"},
 ]
 
 
@@ -185,7 +185,7 @@ def calc_timeout_duration(model_name, text, image, max_new_tokens, temperature,
 
 @spaces.GPU(duration=calc_timeout_duration)
 def generate_image(model_name, text, image, max_new_tokens, temperature, top_p, top_k, repetition_penalty, gpu_timeout):
-    if model_name not in MODEL_MAP:
+    if not model_name or model_name not in MODEL_MAP:
         raise gr.Error("Please select a valid model.")
     if image is None:
         raise gr.Error("Please upload an image.")
@@ -239,35 +239,6 @@ def generate_image(model_name, text, image, max_new_tokens, temperature, top_p,
         torch.cuda.empty_cache()
 
 
-def b64_to_pil(b64_str):
-    if not b64_str:
-        return None
-    try:
-        if b64_str.startswith("data:image"):
-            _, data = b64_str.split(",", 1)
-        else:
-            data = b64_str
-        image_data = base64.b64decode(data)
-        return Image.open(BytesIO(image_data)).convert("RGB")
-    except Exception:
-        return None
-
-
-def run_ocr(model_name, text, image_b64, max_new_tokens_v, temperature_v, top_p_v, top_k_v, repetition_penalty_v, gpu_timeout_v):
-    image = b64_to_pil(image_b64)
-    yield from generate_image(
-        model_name=model_name,
-        text=text,
-        image=image,
-        max_new_tokens=max_new_tokens_v,
-        temperature=temperature_v,
-        top_p=top_p_v,
-        top_k=top_k_v,
-        repetition_penalty=repetition_penalty_v,
-        gpu_timeout=gpu_timeout_v,
-    )
-
-
 def noop():
     return None
 
@@ -302,7 +273,7 @@ footer{display:none!important}
 }
 .app-header-left{display:flex;align-items:center;gap:12px}
 .app-logo{
-    width:38px;height:38px;background:linear-gradient(135deg,#ADFF2F,#C9FF66,#DFFF93);
+    width:38px;height:38px;background:linear-gradient(135deg,#ADFF2F,#C6FF66,#D8FF8A);
     border-radius:10px;display:flex;align-items:center;justify-content:center;
     box-shadow:0 4px 12px rgba(173,255,47,.28);
 }
@@ -313,9 +284,9 @@ footer{display:none!important}
 }
 .app-badge{
     font-size:11px;font-weight:600;padding:3px 10px;border-radius:20px;
-    background:rgba(173,255,47,.10);color:#D6FFA0;border:1px solid rgba(173,255,47,.25);letter-spacing:.3px;
+    background:rgba(173,255,47,.12);color:#D6FF8C;border:1px solid rgba(173,255,47,.28);letter-spacing:.3px;
 }
-.app-badge.fast{background:rgba(173,255,47,.08);color:#C9FF66;border:1px solid rgba(173,255,47,.22)}
+.app-badge.fast{background:rgba(173,255,47,.08);color:#C6FF66;border:1px solid rgba(173,255,47,.22)}
 
 .model-tabs-bar{
     background:#18181b;border-bottom:1px solid #27272a;padding:10px 16px;
@@ -327,9 +298,8 @@ footer{display:none!important}
     border-radius:999px;cursor:pointer;font-size:12px;font-weight:600;padding:0 12px;
     color:#ffffff!important;transition:all .15s ease;
 }
-.model-tab:hover{background:rgba(173,255,47,.12);border-color:rgba(173,255,47,.35)}
-.model-tab.active{background:rgba(173,255,47,.20);border-color:#ADFF2F;color:#111!important;box-shadow:0 0 0 2px rgba(173,255,47,.10)}
-.model-tab.active .model-tab-label{color:#111!important}
+.model-tab:hover{background:rgba(173,255,47,.10);border-color:rgba(173,255,47,.35)}
+.model-tab.active{background:rgba(173,255,47,.18);border-color:#ADFF2F;color:#fff!important;box-shadow:0 0 0 2px rgba(173,255,47,.08)}
 .model-tab-label{font-size:12px;color:#ffffff!important;font-weight:600}
 
 .app-main-row{display:flex;gap:0;flex:1;overflow:hidden}
@@ -383,10 +353,10 @@ footer{display:none!important}
 .preview-action-btn:hover{background:#ADFF2F;border-color:#ADFF2F;color:#111!important}
 
 .hint-bar{
-    background:rgba(173,255,47,.06);border-top:1px solid #27272a;border-bottom:1px solid #27272a;
+    background:rgba(173,255,47,.05);border-top:1px solid #27272a;border-bottom:1px solid #27272a;
     padding:10px 20px;font-size:13px;color:#a1a1aa;line-height:1.7;
 }
-.hint-bar b{color:#D6FFA0;font-weight:600}
+.hint-bar b{color:#D6FF8C;font-weight:600}
 .hint-bar kbd{
     display:inline-block;padding:1px 6px;background:#27272a;border:1px solid #3f3f46;
     border-radius:4px;font-family:'JetBrains Mono',monospace;font-size:11px;color:#a1a1aa;
@@ -406,7 +376,7 @@ footer{display:none!important}
     flex-shrink:0;width:220px;background:#09090b;border:1px solid #27272a;
     border-radius:10px;overflow:hidden;cursor:pointer;transition:all .2s ease;
 }
-.example-card:hover{border-color:#ADFF2F;transform:translateY(-2px);box-shadow:0 4px 12px rgba(173,255,47,.14)}
+.example-card:hover{border-color:#ADFF2F;transform:translateY(-2px);box-shadow:0 4px 12px rgba(173,255,47,.12)}
 .example-card.loading{opacity:.5;pointer-events:none}
 .example-thumb-wrap{height:120px;overflow:hidden;background:#18181b}
 .example-thumb-wrap img{width:100%;height:100%;object-fit:cover}
@@ -417,7 +387,7 @@ footer{display:none!important}
 .example-meta-row{padding:6px 10px;display:flex;align-items:center;gap:6px}
 .example-badge{
     display:inline-flex;padding:2px 7px;background:rgba(173,255,47,.12);border-radius:4px;
-    font-size:10px;font-weight:600;color:#D6FFA0;font-family:'JetBrains Mono',monospace;white-space:nowrap;
+    font-size:10px;font-weight:600;color:#D6FF8C;font-family:'JetBrains Mono',monospace;white-space:nowrap;
 }
 .example-prompt-text{
     padding:0 10px 8px;font-size:11px;color:#a1a1aa;line-height:1.4;
@@ -436,7 +406,7 @@ footer{display:none!important}
     padding:10px 14px;font-family:'Inter',sans-serif;font-size:14px;color:#e4e4e7;
     resize:none;outline:none;min-height:100px;transition:border-color .2s;
 }
-.modern-textarea:focus{border-color:#ADFF2F;box-shadow:0 0 0 3px rgba(173,255,47,.15)}
+.modern-textarea:focus{border-color:#ADFF2F;box-shadow:0 0 0 3px rgba(173,255,47,.14)}
 .modern-textarea::placeholder{color:#3f3f46}
 .modern-textarea.error-flash{
     border-color:#ef4444!important;box-shadow:0 0 0 3px rgba(239,68,68,.2)!important;animation:shake .4s ease;
@@ -452,23 +422,23 @@ footer{display:none!important}
 }
 .toast-notification.visible{transform:translateX(-50%) translateY(0);opacity:1;pointer-events:auto}
 .toast-notification.error{background:linear-gradient(135deg,#dc2626,#b91c1c);color:#fff;border:1px solid rgba(255,255,255,.15)}
-.toast-notification.warning{background:linear-gradient(135deg,#d97706,#b45309);color:#fff;border:1px solid rgba(255,255,255,.15)}
-.toast-notification.info{background:linear-gradient(135deg,#65a30d,#4d7c0f);color:#fff;border:1px solid rgba(255,255,255,.15)}
+.toast-notification.warning{background:linear-gradient(135deg,#84cc16,#65a30d);color:#111;border:1px solid rgba(255,255,255,.08)}
+.toast-notification.info{background:linear-gradient(135deg,#a3e635,#84cc16);color:#111;border:1px solid rgba(255,255,255,.08)}
 .toast-notification .toast-icon{font-size:16px;line-height:1}
 .toast-notification .toast-text{line-height:1.3}
 
 .btn-run{
     display:flex;align-items:center;justify-content:center;gap:8px;width:100%;
-    background:linear-gradient(135deg,#ADFF2F,#95DE2B);border:none;border-radius:10px;
+    background:linear-gradient(135deg,#ADFF2F,#8FD61F);border:none;border-radius:10px;
     padding:12px 24px;cursor:pointer;font-size:15px;font-weight:700;font-family:'Inter',sans-serif;
     color:#111!important;transition:all .2s ease;letter-spacing:-.2px;
-    box-shadow:0 4px 16px rgba(173,255,47,.24),inset 0 1px 0 rgba(255,255,255,.15);
+    box-shadow:0 4px 16px rgba(173,255,47,.25),inset 0 1px 0 rgba(255,255,255,.25);
 }
 .btn-run:hover{
-    background:linear-gradient(135deg,#C9FF66,#ADFF2F);transform:translateY(-1px);
-    box-shadow:0 6px 24px rgba(173,255,47,.34),inset 0 1px 0 rgba(255,255,255,.2);
+    background:linear-gradient(135deg,#C6FF66,#ADFF2F);transform:translateY(-1px);
+    box-shadow:0 6px 24px rgba(173,255,47,.35),inset 0 1px 0 rgba(255,255,255,.25);
 }
-.btn-run:active{transform:translateY(0);box-shadow:0 2px 8px rgba(173,255,47,.24)}
+.btn-run:active{transform:translateY(0);box-shadow:0 2px 8px rgba(173,255,47,.25)}
 .btn-run svg{width:18px;height:18px;fill:#111!important}
 .btn-run svg path{fill:#111!important}
 
@@ -480,12 +450,12 @@ footer{display:none!important}
 }
 .out-title-right{display:flex;gap:8px;align-items:center}
 .out-action-btn{
-    display:inline-flex;align-items:center;justify-content:center;background:rgba(173,255,47,.1);
+    display:inline-flex;align-items:center;justify-content:center;background:rgba(173,255,47,.10);
     border:1px solid rgba(173,255,47,.2);border-radius:6px;cursor:pointer;padding:3px 10px;
-    font-size:11px;font-weight:500;color:#D6FFA0!important;gap:4px;height:24px;transition:all .15s;
+    font-size:11px;font-weight:500;color:#D6FF8C!important;gap:4px;height:24px;transition:all .15s;
 }
 .out-action-btn:hover{background:rgba(173,255,47,.2);border-color:rgba(173,255,47,.35);color:#111!important}
-.out-action-btn svg{width:12px;height:12px;fill:#D6FFA0}
+.out-action-btn svg{width:12px;height:12px;fill:#D6FF8C}
 .output-frame .out-body{
     flex:1;background:#09090b;display:flex;align-items:stretch;justify-content:stretch;
     overflow:hidden;min-height:320px;position:relative;
@@ -515,7 +485,7 @@ footer{display:none!important}
 .modern-loader .loader-text{font-size:13px;color:#a1a1aa;font-weight:500}
 .loader-bar-track{width:200px;height:4px;background:#27272a;border-radius:2px;overflow:hidden}
 .loader-bar-fill{
-    height:100%;background:linear-gradient(90deg,#ADFF2F,#C9FF66,#ADFF2F);
+    height:100%;background:linear-gradient(90deg,#ADFF2F,#C6FF66,#ADFF2F);
     background-size:200% 100%;animation:shimmer 1.5s ease-in-out infinite;border-radius:2px;
 }
 @keyframes shimmer{0%{background-position:200% 0}100%{background-position:-200% 0}}
@@ -533,12 +503,12 @@ footer{display:none!important}
     border-radius:3px;outline:none;min-width:0;
 }
 .slider-row input[type="range"]::-webkit-slider-thumb{
-    -webkit-appearance:none;width:16px;height:16px;background:linear-gradient(135deg,#ADFF2F,#95DE2B);
+    -webkit-appearance:none;width:16px;height:16px;background:linear-gradient(135deg,#ADFF2F,#8FD61F);
     border-radius:50%;cursor:pointer;box-shadow:0 2px 6px rgba(173,255,47,.35);transition:transform .15s;
 }
 .slider-row input[type="range"]::-webkit-slider-thumb:hover{transform:scale(1.2)}
 .slider-row input[type="range"]::-moz-range-thumb{
-    width:16px;height:16px;background:linear-gradient(135deg,#ADFF2F,#95DE2B);
+    width:16px;height:16px;background:linear-gradient(135deg,#ADFF2F,#8FD61F);
     border-radius:50%;cursor:pointer;border:none;box-shadow:0 2px 6px rgba(173,255,47,.35);
 }
 .slider-row .slider-val{
@@ -557,11 +527,11 @@ footer{display:none!important}
 }
 .app-statusbar .sb-section.sb-fixed{
     flex:0 0 auto;min-width:110px;text-align:center;justify-content:center;
-    padding:3px 12px;background:rgba(173,255,47,.08);border-radius:6px;color:#D6FFA0;font-weight:500;
+    padding:3px 12px;background:rgba(173,255,47,.08);border-radius:6px;color:#D6FF8C;font-weight:500;
 }
 
 .exp-note{padding:10px 20px;font-size:12px;color:#52525b;border-top:1px solid #27272a;text-align:center}
-.exp-note a{color:#D6FFA0;text-decoration:none}
+.exp-note a{color:#D6FF8C;text-decoration:none}
 .exp-note a:hover{text-decoration:underline}
 
 ::-webkit-scrollbar{width:8px;height:8px}
@@ -579,7 +549,7 @@ footer{display:none!important}
 gallery_js = r"""
 () => {
 function init() {
-    if (window.__ocrGreenInitDone) return;
+    if (window.__ocr2GreenInitDone) return;
 
     const dropZone = document.getElementById('image-drop-zone');
     const uploadPrompt = document.getElementById('upload-prompt');
@@ -600,7 +570,7 @@ function init() {
         return;
     }
 
-    window.__ocrGreenInitDone = true;
+    window.__ocr2GreenInitDone = true;
     let imageState = null;
     let toastTimer = null;
 
@@ -887,11 +857,11 @@ function init() {
         if (!exampleResultContainer) return;
         const el = exampleResultContainer.querySelector('textarea') || exampleResultContainer.querySelector('input');
         if (!el || !el.value) return;
-        if (window.__lastExampleVal === el.value) return;
+        if (window.__lastExampleVal2 === el.value) return;
         try {
             const data = JSON.parse(el.value);
             if (data.status === 'ok') {
-                window.__lastExampleVal = el.value;
+                window.__lastExampleVal2 = el.value;
                 if (data.image) setPreview(data.image, data.name || 'example.jpg');
                 if (data.query) {
                     promptInput.value = data.query;
@@ -965,7 +935,7 @@ UPLOAD_PREVIEW_SVG = """
 <svg viewBox="0 0 80 80" fill="none" xmlns="http://www.w3.org/2000/svg">
     <rect x="8" y="14" width="64" height="52" rx="6" fill="none" stroke="#ADFF2F" stroke-width="2" stroke-dasharray="4 3"/>
     <polygon points="12,62 30,40 42,50 54,34 68,62" fill="rgba(173,255,47,0.15)" stroke="#ADFF2F" stroke-width="1.5"/>
-    <circle cx="28" cy="30" r="6" fill="rgba(173,255,47,0.2)" stroke="#ADFF2F" stroke-width="1.5"/>
+    <circle cx="28" cy="30" r="6" fill="rgba(173,255,47,0.22)" stroke="#ADFF2F" stroke-width="1.5"/>
 </svg>
 """
 
@@ -1054,7 +1024,7 @@ with gr.Blocks() as demo:
                     <div class="panel-card-title">OCR / Vision Instruction</div>
                     <div class="panel-card-body">
                         <label class="modern-label" for="custom-query-input">Query Input</label>
-                        <textarea id="custom-query-input" class="modern-textarea" rows="4" placeholder="e.g., perform OCR on the image precisely, extract all text, read the document, summarize visual content, describe the image with text..."></textarea>
+                        <textarea id="custom-query-input" class="modern-textarea" rows="4" placeholder="e.g., perform OCR on the image precisely, extract all text, describe the visual scene, summarize visible content..."></textarea>
                     </div>
                 </div>
 
@@ -1124,7 +1094,7 @@ with gr.Blocks() as demo:
         </div>
 
         <div class="exp-note">
-            Experimental OCR Suite &middot; Open on <a href="https://github.com/PRITHIVSAKTHIUR/Multimodal-OCR" target="_blank">GitHub</a>
+            Experimental OCR Suite &middot; Open on <a href="https://github.com/PRITHIVSAKTHIUR/Multimodal-OCR2" target="_blank">GitHub</a>
         </div>
 
         <div class="app-statusbar">
@@ -1136,6 +1106,33 @@ with gr.Blocks() as demo:
 
     run_btn = gr.Button("Run", elem_id="gradio-run-btn")
 
+    def b64_to_pil(b64_str):
+        if not b64_str:
+            return None
+        try:
+            if b64_str.startswith("data:image"):
+                _, data = b64_str.split(",", 1)
+            else:
+                data = b64_str
+            image_data = base64.b64decode(data)
+            return Image.open(BytesIO(image_data)).convert("RGB")
+        except Exception:
+            return None
+
+    def run_ocr(model_name, text, image_b64, max_new_tokens_v, temperature_v, top_p_v, top_k_v, repetition_penalty_v, gpu_timeout_v):
+        image = b64_to_pil(image_b64)
+        yield from generate_image(
+            model_name=model_name,
+            text=text,
+            image=image,
+            max_new_tokens=max_new_tokens_v,
+            temperature=temperature_v,
+            top_p=top_p_v,
+            top_k=top_k_v,
+            repetition_penalty=repetition_penalty_v,
+            gpu_timeout=gpu_timeout_v,
+        )
+
     demo.load(fn=noop, inputs=None, outputs=None, js=gallery_js)
     demo.load(fn=noop, inputs=None, outputs=None, js=wire_outputs_js)