import streamlit as st import os """ # Welcome to Streamlit! Edit `/streamlit_app.py` to customize this app to your heart's desire :heart:. If you have any questions, checkout our [documentation](https://docs.streamlit.io) and [community forums](https://discuss.streamlit.io). In the meantime, below is an example of what you can do with just a few lines of code: """ # ✅ MUST be the very first Streamlit call st.set_page_config( page_title="SmartVision", layout="wide", page_icon="👁️" ) # ───────────────────────────────────────────────────────────── # HERO WITH IMAGE # ───────────────────────────────────────────────────────────── st.markdown( """

""", unsafe_allow_html=True, ) # ✅ FIX 1: Use forward slash (works on Linux/Mac/Windows) # image_path = "models/smartvision.png" import os APP_ROOT = "/app" image_path = os.path.join(APP_ROOT, "models", "smartvision.png") # ✅ FIX 2: Check if file exists before loading to give a clear error if os.path.exists(image_path): st.image(image_path, use_container_width=True) else: st.warning(f"⚠️ Image not found at `{image_path}`. Please ensure the file is committed to your repository.") st.markdown( """

Unified Image Classification & Object Detection Platform

25-Class COCO · YOLOv8 · VGG16 · ResNet50 · MobileNetV2 · EfficientNetB0

""", unsafe_allow_html=True, ) st.info("📌 Use the **sidebar** to navigate between pages.") # ───────────────────────────────────────────────────────────────────────────── # SECTION 1 — TOP STATS # ───────────────────────────────────────────────────────────────────────────── st.markdown("### 📊 Platform Overview") c1, c2, c3, c4, c5 = st.columns(5) stats = [ ("🎯 Classes", "25", "#2ecc71"), ("🧠 CNN Models", "4", "#3498db"), ("🔍 Detection", "YOLOv8s", "#f39c12"), ("📦 Dataset", "COCO", "#9b59b6"), ("☁️ Deployment", "HF Spaces", "#1abc9c"), ] for col, (label, val, color) in zip([c1, c2, c3, c4, c5], stats): with col: st.markdown( f"""

{label}

{val}

""", unsafe_allow_html=True, ) st.divider() # ───────────────────────────────────────────────────────────────────────────── # SECTION 2 — KEY FEATURES # ───────────────────────────────────────────────────────────────────────────── st.markdown("### 🚀 Key Features") feat_col1, feat_col2 = st.columns(2) features_left = [ ("🖼️", "Single-object image classification", "Upload any image — all 4 CNNs classify it simultaneously with Top-5 predictions."), ("📦", "Multi-object detection with bounding boxes","YOLOv8 detects and localises multiple objects in a single frame."), ("🎨", "Color-coded per-class visualisation", "Each of 25 classes has a unique bounding box color for instant recognition."), ] features_right = [ ("📊", "Interactive performance dashboard", "Live charts, confusion matrices, training curves, and per-class mAP breakdown."), ("📷", "Live webcam detection", "Real-time YOLOv8 inference on webcam feed with FPS overlay (local only)."), ] for col, feats in zip([feat_col1, feat_col2], [features_left, features_right]): with col: for icon, title, desc in feats: st.markdown( f"""

{icon}

{title}

{desc}

""", unsafe_allow_html=True, ) st.divider() # ===================================================== # QUICK NAVIGATION BUTTONS # ===================================================== st.header("🚀 Explore Features") col1, col2 = st.columns(2) with col1: if st.button("🖼️ Image Classification", use_container_width=True): st.switch_page("pages/image_classification.py") if st.button("📈 CNN Performance", use_container_width=True): st.switch_page("pages/model_performance.py") with col2: if st.button("🔍 Object Detection", use_container_width=True): st.switch_page("pages/yolo_object_detection.py") if st.button("📊 YOLO Performance", use_container_width=True): st.switch_page("pages/yolo_performance.py") st.divider() # ───────────────────────────────────────────────────────────────────────────── # SECTION 3 — MODELS USED # ───────────────────────────────────────────────────────────────────────────── st.markdown("### 🤖 Models Used") # ── YOLOv8 ─────────────────────────────────────────────────────────────────── st.markdown("#### 🔍 Object Detection") yolo_col, yolo_info = st.columns([1, 3]) with yolo_col: st.markdown( """

⚡

YOLOv8s

Ultralytics

""", unsafe_allow_html=True, ) with yolo_info: chips = [ ("ARCHITECTURE", "YOLOv8 Small (YOLOv8s)"), ("PRETRAINED ON", "COCO (80 classes)"), ("FINE-TUNED ON", "25-Class COCO Subset"), ("EPOCHS", "50"), ("IMAGE SIZE", "640 × 640"), ("mAP@0.5", "0.893"), ("PRECISION", "95.1%"), ("RECALL", "87.7%"), ("INFERENCE", "~6.7 ms · ~120 FPS"), ] chips_html = "".join( f'' f'{k}' f'{v}' for k, v in chips ) st.markdown( f"""

{chips_html}

YOLOv8s is Ultralytics' small variant — optimised for the best balance between accuracy and inference speed. Fine-tuned with mosaic augmentation, auto-optimizer (AdamW), and AMP mixed precision.

""", unsafe_allow_html=True, ) st.markdown("
", unsafe_allow_html=True) # ── CNN Models ───────────────────────────────────────────── st.markdown("#### 🧠 Image Classification — CNN Models") CNN_MODELS = [ {"name": "EfficientNetB0", "icon": "🟨", "color": "#f39c12", "highlight": "Best Accuracy"}, {"name": "ResNet50", "icon": "🟥", "color": "#e74c3c", "highlight": "Best All-Rounder"}, {"name": "MobileNetV2", "icon": "🟩", "color": "#2ecc71", "highlight": "Fastest Inference"}, {"name": "VGG16", "icon": "🟦", "color": "#3498db", "highlight": "Reliable Baseline"}, ] for model in CNN_MODELS: st.markdown( f"""

{model['icon']} {model['name']} {model['highlight']}

""", unsafe_allow_html=True ) st.divider() # ───────────────────────────────────────────────────────────────────────────── # SECTION 4 — DATASET + CLASSES # ───────────────────────────────────────────────────────────────────────────── st.markdown("### 📦 Dataset") d1, d2 = st.columns([1, 2]) with d1: dataset_rows = [ ("Total Images", "2,500"), ("Images / Class", "100"), ("Train Split", "70% · 1,750 images"), ("Val Split", "15% · 375 images"), ("Test Split", "15% · 375 images"), ("Image Size", "224 × 224 px"), ("Annotation", "COCO Bounding Boxes"), ("Source", "Hugging Face (streaming)"), ] rows_html = "".join( f'

' f'{k}' f'{v}

' for k, v in dataset_rows ) st.markdown( f'

' f'

COCO-25 Subset

' f'{rows_html}

', unsafe_allow_html=True, ) with d2: st.markdown("**25 Supported Classes:**") CLASS_NAMES = [ 'person','bicycle','car','motorcycle','airplane','bus','train','truck', 'traffic light','stop sign','bench','bird','cat','dog','horse','cow', 'elephant','bottle','cup','bowl','pizza','cake','chair','couch','potted plant', ] ICONS = { 'person':'🧍','bicycle':'🚲','car':'🚗','motorcycle':'🏍️','airplane':'✈️', 'bus':'🚌','train':'🚆','truck':'🚛','traffic light':'🚦','stop sign':'🛑', 'bench':'🪑','bird':'🐦','cat':'🐱','dog':'🐶','horse':'🐴','cow':'🐮', 'elephant':'🐘','bottle':'🍶','cup':'☕','bowl':'🥣','pizza':'🍕', 'cake':'🎂','chair':'🪑','couch':'🛋️','potted plant':'🪴', } badges = " ".join( f'' f'{ICONS.get(c,"•")} {c}' for c in CLASS_NAMES ) st.markdown(f'

{badges}

', unsafe_allow_html=True) st.divider() # ───────────────────────────────────────────────────────────── # SECTION 5 — BUSINESS APPLICATIONS # ───────────────────────────────────────────────────────────── st.markdown("### 🌍 Business Applications") APPS = [ ("🏙️", "Smart Cities & Traffic Management", "Automated vehicle detection · Pedestrian safety monitoring · Parking detection · Traffic violation detection"), ("🛒", "Retail & E-Commerce", "Product recognition · Scan-free checkout · Customer behaviour analytics · Visual search"), ("🔒", "Security & Surveillance", "Intrusion detection · Unattended object alerts · Perimeter monitoring · Crowd density analysis"), ("🦁", "Wildlife Conservation", "Species identification · Habitat monitoring · Poaching prevention · Population studies"), ("🏥", "Healthcare", "PPE compliance verification · Equipment tracking · Patient fall detection · Hygiene monitoring"), ("🏠", "Smart Home & IoT", "Home automation · Security alerts · Pet activity tracking · Energy usage detection"), ("🌾", "Agriculture", "Livestock monitoring · Pest detection · Equipment tracking · Harvest readiness detection"), ("📦", "Logistics & Warehousing", "Package sorting · Inventory tracking · Quality control · Loading bay monitoring"), ] for icon, title, description in APPS: st.markdown( f"""

{icon} {title}: {description}

""", unsafe_allow_html=True, ) st.divider() # ───────────────────────────────────────────────────────────── # SECTION 6 — TECH STACK + RESULTS # ───────────────────────────────────────────────────────────── st.markdown("### 🛠️ Tech Stack & Project Info") tech_col, dev_col = st.columns([1, 1]) with tech_col: TECH = [ ("🐍", "Python 3.10", "Core programming language"), ("🤖", "TensorFlow / Keras", "CNN training & inference"), ("⚡", "Ultralytics YOLOv8", "Object detection & fine-tuning"), ("👁️", "OpenCV", "Image processing & annotation"), ("🌐", "Streamlit", "Interactive web app framework"), ("🤗", "Hugging Face Spaces", "Cloud deployment platform"), ("📊", "Matplotlib / Seaborn", "Visualisations"), ("🔢", "NumPy / Pandas", "Data processing"), ("🧪", "Scikit-learn", "Evaluation metrics"), ] rows_html = "".join( f'

' f'{icon}' f'{name}' f'{desc}

' for icon, name, desc in TECH ) st.markdown( f'''

🛠️ Tech Stack

{rows_html}

''', unsafe_allow_html=True, ) with dev_col: st.markdown("""

📊 Model Scope & Results

Image Classification (Transfer Learning)
Object Detection (YOLOv8 Fine-tuning)
Evaluation — Accuracy · mAP@0.5 · Precision · Recall · F1

CNN Result — 92% (EfficientNetB0) ✅
YOLO Result — mAP@0.5 = 0.893 ✅

""", unsafe_allow_html=True)