Spaces:
Running
Running
| title: Tokiva Product Detector | |
| emoji: 🛒 | |
| colorFrom: blue | |
| colorTo: purple | |
| sdk: static | |
| pinned: false | |
| # 🛒 Tokiva Product Detector | |
| Aplikasi web untuk mendeteksi produk toko menggunakan model **MobileNetV4 + CBAM** (akurasi 98.77%). | |
| ## Fitur | |
| - 📸 **Deteksi via Webcam** — Scan produk langsung dari kamera | |
| - 🖼️ **Upload Foto** — Upload gambar produk untuk dideteksi | |
| - 📊 **Batch Test** — Test banyak gambar sekaligus + export CSV | |
| - 🔬 **Analisis Detail** — GradCAM heatmap + probability chart | |
| ## Model Info | |
| | Property | Value | | |
| |---|---| | |
| | Architecture | MobileNetV4 + CBAM | | |
| | Accuracy | 98.77% | | |
| | F1 Score | 0.9897 | | |
| | Classes | 19 produk | | |
| | Model Size | ~12 MB | | |
| ## Produk yang Dikenali | |
| Frisian Flag, Gaga Mie, Indomie, Klik Crackers, Nabati Siip, Pepsodent, Sarimi, Sedaap, Siplah, Soffell, Ultra Milk, Vica. | |
| ## Tech Stack | |
| - **Model**: TensorFlow (SavedModel format) | |
| - **UI**: Gradio | |
| - **Deploy**: HuggingFace Spaces | |