prithivMLmods/Weather-Image-Classification

1---2license: apache-2.03datasets:4- prithivMLmods/WeatherNet-055library_name: transformers6language:7- en8base_model:9- google/siglip2-base-patch16-22410pipeline_tag: image-classification11tags:12- Weather-Detection13- SigLIP214- 93M15---16 17![1.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/DLSG05GqVrEJR7dE3VoiV.png)18 19# Weather-Image-Classification20 21> Weather-Image-Classification is a vision-language model fine-tuned from google/siglip2-base-patch16-224 for multi-class image classification. It is trained to recognize weather conditions from images using the SiglipForImageClassification architecture.22 23```py24Classification Report:25                 precision    recall  f1-score   support26 27cloudy/overcast     0.8493    0.8762    0.8625      670228     foggy/hazy     0.8340    0.8128    0.8233      126129     rain/strom     0.7644    0.7592    0.7618      192730    snow/frosty     0.8341    0.8448    0.8394      187531      sun/clear     0.9124    0.8846    0.8983      627432 33       accuracy                         0.8589     1803934      macro avg     0.8388    0.8355    0.8371     1803935   weighted avg     0.8595    0.8589    0.8591     1803936```37 38![download (1).png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/T3MuycHMZDoAhjp3V5Z0p.png)39 40---41 42## Label Space: 5 Classes43 44The model classifies an image into one of the following weather categories:45 46```json47"id2label": {48  "0": "cloudy/overcast",49  "1": "foggy/hazy",50  "2": "rain/storm",51  "3": "snow/frosty",52  "4": "sun/clear"53}54```55 56---57 58## Install Dependencies59 60```bash61pip install -q transformers torch pillow gradio62```63 64---65 66## Inference Code67 68```python69import gradio as gr70from transformers import AutoImageProcessor, SiglipForImageClassification71from PIL import Image72import torch73 74# Load model and processor75model_name = "prithivMLmods/Weather-Image-Classification"  # Replace with actual path76model = SiglipForImageClassification.from_pretrained(model_name)77processor = AutoImageProcessor.from_pretrained(model_name)78 79# Label mapping80id2label = {81    "0": "cloudy/overcast",82    "1": "foggy/hazy",83    "2": "rain/storm",84    "3": "snow/frosty",85    "4": "sun/clear"86}87 88def classify_weather(image):89    image = Image.fromarray(image).convert("RGB")90    inputs = processor(images=image, return_tensors="pt")91 92    with torch.no_grad():93        outputs = model(**inputs)94        logits = outputs.logits95        probs = torch.nn.functional.softmax(logits, dim=1).squeeze().tolist()96 97    prediction = {98        id2label[str(i)]: round(probs[i], 3) for i in range(len(probs))99    }100 101    return prediction102 103# Gradio Interface104iface = gr.Interface(105    fn=classify_weather,106    inputs=gr.Image(type="numpy"),107    outputs=gr.Label(num_top_classes=5, label="Weather Condition"),108    title="Weather-Image-Classification",109    description="Upload an image to identify the weather condition (sun, rain, snow, fog, or clouds)."110)111 112if __name__ == "__main__":113    iface.launch()114```115 116---117 118## Intended Use119 120Weather-Image-Classification is useful for:121 122* Automated weather tagging for photography and media.123* Enhancing dataset labeling in weather-related research.124* Supporting smart surveillance and traffic systems.125* Improving scene understanding in autonomous vehicles.