SigLIP2 Content Filters - Models v.1
Collection
Moderation, Balance, Classifiers
β’
15 items
β’
Updated
β’
2
Recycling-Net-11 is an image classification model fine-tuned from google/siglip2-base-patch16-224 using the SiglipForImageClassification architecture. The model classifies images into 11 categories related to recyclable materials, helping to automate and enhance waste sorting systems.
Classification Report:
precision recall f1-score support
aluminium 0.9213 0.9145 0.9179 269
batteries 0.9833 0.9933 0.9883 297
cardboard 0.9660 0.9343 0.9499 274
disposable plates 0.9078 0.9744 0.9399 273
glass 0.9621 0.9490 0.9555 294
hard plastic 0.8675 0.7250 0.7899 280
paper 0.8702 0.8941 0.8820 255
paper towel 0.9333 0.9622 0.9475 291
polystyrene 0.8188 0.8385 0.8285 291
soft plastics 0.8425 0.8693 0.8557 283
takeaway cups 0.9575 0.9767 0.9670 300
accuracy 0.9128 3107
macro avg 0.9119 0.9119 0.9111 3107
weighted avg 0.9127 0.9128 0.9119 3107
The model categorizes images into the following classes:
!pip install -q transformers torch pillow gradio
import gradio as gr
from transformers import AutoImageProcessor, SiglipForImageClassification
from PIL import Image
import torch
# Load model and processor
model_name = "prithivMLmods/Recycling-Net-11" # Update with your actual Hugging Face model path
model = SiglipForImageClassification.from_pretrained(model_name)
processor = AutoImageProcessor.from_pretrained(model_name)
# Label mapping
id2label = {
0: "aluminium",
1: "batteries",
2: "cardboard",
3: "disposable plates",
4: "glass",
5: "hard plastic",
6: "paper",
7: "paper towel",
8: "polystyrene",
9: "soft plastics",
10: "takeaway cups"
}
def classify_recyclable_material(image):
"""Predicts the type of recyclable material in the image."""
image = Image.fromarray(image).convert("RGB")
inputs = processor(images=image, return_tensors="pt")
with torch.no_grad():
outputs = model(**inputs)
logits = outputs.logits
probs = torch.nn.functional.softmax(logits, dim=1).squeeze().tolist()
predictions = {id2label[i]: round(probs[i], 3) for i in range(len(probs))}
return predictions
# Gradio interface
iface = gr.Interface(
fn=classify_recyclable_material,
inputs=gr.Image(type="numpy"),
outputs=gr.Label(label="Recyclable Material Prediction Scores"),
title="Recycling-Net-11",
description="Upload an image of a waste item to identify its recyclable material type."
)
# Launch the app
if __name__ == "__main__":
iface.launch()
Recycling-Net-11 is ideal for:
Base model
google/siglip2-so400m-patch14-384