metadata
library_name: pytorch
license: apache-2.0
pipeline_tag: image-classification
tags:
- mobilenetv3
- image-classification
- fruit-quality
- torchvision
datasets:
- custom/fruit-quality-dataset
base_model: pytorch/vision-mobilenet_v3_large
Fruit Quality Classifier (MobileNetV3-Large)
Model ini adalah hasil fine-tuning dari MobileNetV3-Large (pretrained ImageNet) untuk klasifikasi kualitas / kondisi buah pisang, jeruk dan apel (fresh atau rotten) menggunakan dataset kustom.
Ringkasan
- Arsitektur: MobileNetV3-Large (torchvision)
- Jumlah kelas: 6 (['freshapples', 'freshbanana', 'freshoranges', 'rottenapples', 'rottenbanana', 'rottenoranges'])
- Input: gambar RGB, ukuran 224×224
- Preprocessing: Resize → ToTensor → Normalize (mean=[0.485,0.456,0.406], std=[0.229,0.224,0.225])
- Augmentasi (saat training): RandomHorizontalFlip, RandomRotation, RandomColorJitter, RandomPerspective, Dropout pada classifier
Hyperparameter
| Parameter | Nilai |
|---|---|
| Model Backbone | MobileNetV3-Large (IMAGENET1K_V1 pretrained) |
| Optimizer | Adam |
| Learning Rate | 1e-4 |
| Weight Decay | 1e-5 |
| Loss Function | CrossEntropyLoss |
| Epochs | 10 |
| Batch Size | 32 |
Hasil Evaluasi
Akurasi = 95% Loss = 0,169