--- license: apache-2.0 language: - en base_model: - google/siglip2-base-patch16-224 pipeline_tag: image-classification library_name: transformers tags: - color - cloth --- ![18.png](https://cdn-uploads.huggingface.co/production/uploads/65bb837dbfb878f46c77de4c/B0YuDqYBhhE310ZHAFDFE.png) # **Fashion-Product-baseColour** > **Fashion-Product-baseColour** is a visual classification model fine-tuned from **google/siglip2-base-patch16-224** using the **SiglipForImageClassification** architecture. It predicts the **base color** of fashion products from images — enabling accurate tagging, search, and recommendation in fashion-related applications. ```py Classification Report: precision recall f1-score support Beige 0.4338 0.5409 0.4815 745 Black 0.8051 0.8656 0.8342 9699 Blue 0.7513 0.7858 0.7682 4906 Bronze 0.0000 0.0000 0.0000 89 Brown 0.6812 0.7596 0.7183 3440 Burgundy 0.0000 0.0000 0.0000 44 Charcoal 0.4941 0.1842 0.2684 228 Coffee Brown 0.0000 0.0000 0.0000 29 Copper 0.5000 0.0120 0.0235 83 Cream 0.3940 0.3446 0.3677 383 Fluorescent Green 0.0000 0.0000 0.0000 5 Gold 0.4935 0.6747 0.5701 621 Green 0.7286 0.7760 0.7516 2103 Grey 0.6313 0.5002 0.5581 2735 Grey Melange 0.5728 0.4041 0.4739 146 Khaki 0.3540 0.2878 0.3175 139 Lavender 0.5049 0.3250 0.3954 160 Lime Green 0.0000 0.0000 0.0000 5 Magenta 0.5909 0.1016 0.1733 128 Maroon 0.5121 0.2929 0.3727 577 Mauve 0.0000 0.0000 0.0000 28 Metallic 0.0000 0.0000 0.0000 41 Multi 0.4005 0.3832 0.3917 394 Mushroom Brown 0.0000 0.0000 0.0000 16 Mustard 0.4912 0.2887 0.3636 97 Navy Blue 0.6290 0.4905 0.5512 1784 Nude 0.0000 0.0000 0.0000 21 Off White 0.5789 0.2418 0.3411 182 Olive 0.5259 0.5208 0.5233 409 Orange 0.6838 0.6119 0.6458 523 Peach 0.4727 0.4216 0.4457 185 Pink 0.6912 0.7423 0.7158 1824 Purple 0.6846 0.7568 0.7189 1612 Red 0.6916 0.8273 0.7534 2432 Rose 0.0000 0.0000 0.0000 21 Rust 0.5000 0.1692 0.2529 65 Sea Green 0.0000 0.0000 0.0000 22 Silver 0.6088 0.4830 0.5387 1089 Skin 0.5479 0.6319 0.5869 163 Steel 0.2857 0.0381 0.0672 315 Tan 0.6667 0.0357 0.0678 112 Taupe 0.0000 0.0000 0.0000 11 Teal 0.4857 0.2857 0.3598 119 Turquoise Blue 0.0000 0.0000 0.0000 69 White 0.7518 0.7950 0.7728 5497 Yellow 0.7714 0.8003 0.7856 776 accuracy 0.7072 44072 macro avg 0.4112 0.3343 0.3469 44072 weighted avg 0.6919 0.7072 0.6935 44072 ``` The model categorizes fashion product images into the following **46 base color classes**: - Beige, Black, Blue, Bronze, Brown, Burgundy, Charcoal, Coffee Brown, Copper, Cream - Fluorescent Green, Gold, Green, Grey, Grey Melange, Khaki, Lavender, Lime Green - Magenta, Maroon, Mauve, Metallic, Multi, Mushroom Brown, Mustard, Navy Blue - Nude, Off White, Olive, Orange, Peach, Pink, Purple, Red, Rose, Rust - Sea Green, Silver, Skin, Steel, Tan, Taupe, Teal, Turquoise Blue, White, Yellow --- # **Run with Transformers 🤗** ```python !pip install -q transformers torch pillow gradio ``` ```python import gradio as gr from transformers import AutoImageProcessor, SiglipForImageClassification from PIL import Image import torch # Load model and processor model_name = "prithivMLmods/Fashion-Product-baseColour" # Replace with actual model path model = SiglipForImageClassification.from_pretrained(model_name) processor = AutoImageProcessor.from_pretrained(model_name) # Label mapping id2label = { 0: "Beige", 1: "Black", 2: "Blue", 3: "Bronze", 4: "Brown", 5: "Burgundy", 6: "Charcoal", 7: "Coffee Brown", 8: "Copper", 9: "Cream", 10: "Fluorescent Green", 11: "Gold", 12: "Green", 13: "Grey", 14: "Grey Melange", 15: "Khaki", 16: "Lavender", 17: "Lime Green", 18: "Magenta", 19: "Maroon", 20: "Mauve", 21: "Metallic", 22: "Multi", 23: "Mushroom Brown", 24: "Mustard", 25: "Navy Blue", 26: "Nude", 27: "Off White", 28: "Olive", 29: "Orange", 30: "Peach", 31: "Pink", 32: "Purple", 33: "Red", 34: "Rose", 35: "Rust", 36: "Sea Green", 37: "Silver", 38: "Skin", 39: "Steel", 40: "Tan", 41: "Taupe", 42: "Teal", 43: "Turquoise Blue", 44: "White", 45: "Yellow" } def classify_base_color(image): """Predicts the base color of a fashion product from an 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_base_color, inputs=gr.Image(type="numpy"), outputs=gr.Label(label="Base Colour Prediction Scores"), title="Fashion-Product-baseColour", description="Upload a fashion product image to detect its primary color (e.g., Red, Black, Cream, Navy Blue, etc.)." ) # Launch the app if __name__ == "__main__": iface.launch() ``` --- # **Intended Use** This model is ideal for: - **E-commerce platforms** for accurate product color labeling - **Fashion search engines** and recommendation systems - **Inventory and catalog automation** - **Fashion analytics and trends tracking** - **Design tools** for color-based sorting and filters