|
--- |
|
language: en |
|
tags: |
|
- image-classification |
|
- pytorch |
|
- resnet |
|
- imagenet |
|
datasets: |
|
- imagenet-1k |
|
metrics: |
|
- accuracy |
|
--- |
|
|
|
# ResNet50 ImageNet Classifier |
|
|
|
This model is a ResNet50 architecture trained on the ImageNet dataset for image classification. |
|
|
|
## Model Description |
|
|
|
- **Model Type:** ResNet50 |
|
- **Task:** Image Classification |
|
- **Training Data:** ImageNet (ILSVRC2012) |
|
- **Number of Parameters:** ~23M |
|
- **Input:** RGB images of size 224x224 |
|
|
|
## Usage |
|
|
|
```python |
|
from transformers import AutoFeatureExtractor, AutoModelForImageClassification |
|
import torch |
|
from PIL import Image |
|
|
|
# Load model and feature extractor |
|
model = AutoModelForImageClassification.from_pretrained("jatingocodeo/ImageNet") |
|
feature_extractor = AutoFeatureExtractor.from_pretrained("jatingocodeo/ImageNet") |
|
|
|
# Prepare image |
|
image = Image.open("path/to/image.jpg") |
|
inputs = feature_extractor(image, return_tensors="pt") |
|
|
|
# Get predictions |
|
with torch.no_grad(): |
|
outputs = model(**inputs) |
|
logits = outputs.logits |
|
predicted_class = logits.argmax(-1).item() |
|
``` |
|
|
|
## Training |
|
|
|
The model was trained on the ImageNet dataset with the following configuration: |
|
- Optimizer: AdamW |
|
- Learning Rate: 0.003 with cosine scheduling |
|
- Batch Size: 256 |
|
- Data Augmentation: RandomResizedCrop, RandomHorizontalFlip, ColorJitter, RandomAffine, RandomPerspective |
|
|
|
## Preprocessing |
|
|
|
The model expects images to be preprocessed as follows: |
|
- Resize shortest edge to 224 |
|
- Center crop to 224x224 |
|
- Normalize with mean [0.485, 0.456, 0.406] and std [0.229, 0.224, 0.225] |
|
|
