Fashion Object Detection Yolos Tiny

🚀 Fashion Object Detection with Yolos-Tiny

This is a fine-tuned object detection model based on hustvl/yolos-tiny, specialized for fashion-related object detection.

🚀 Quick Start

This model is a fine-tuned version of hustvl/yolos-tiny. You can find detailed information about the model in this GitHub repository: fashion-visual-search. Additionally, you can find the fashion image feature extractor model here: yainage90/fashion-image-feature-extractor.

✨ Features

• Fine-tuned Model: Based on hustvl/yolos-tiny, optimized for fashion object detection.
• Multi-dataset Training: Trained using a combination of modanet and fashionpedia datasets.
• Specific Labels: The model can detect the following fashion items: ['bag', 'bottom', 'dress', 'hat', 'shoes', 'outer', 'top'].
• Good Performance: Achieved a best mAP score of 0.697400 at the 96th epoch out of 100 epochs.

📦 Installation

No specific installation steps are provided in the original document.

💻 Usage Examples

Basic Usage

from PIL import Image
import torch
from transformers import  YolosImageProcessor, YolosForObjectDetection

device = 'cpu'
if torch.cuda.is_available():
    device = torch.device('cuda')
elif torch.backends.mps.is_available():
    device = torch.device('mps')

ckpt = 'yainage90/fashion-object-detection-yolos-tiny'
image_processor = YolosImageProcessor.from_pretrained(ckpt)
model = YolosForObjectDetection.from_pretrained(ckpt).to(device)

image = Image.open('<path/to/image>').convert('RGB')

with torch.no_grad():
    inputs = image_processor(images=[image], return_tensors="pt")
    outputs = model(**inputs.to(device))
    target_sizes = torch.tensor([[image.size[1], image.size[0]]])
    results = image_processor.post_process_object_detection(outputs, threshold=0.85, target_sizes=target_sizes)[0]

    items = []
    for score, label, box in zip(results["scores"], results["labels"], results["boxes"]):
        score = score.item()
        label = label.item()
        box = [i.item() for i in box]
        print(f"{model.config.id2label[label]}: {round(score, 3)} at {box}")
        items.append((score, label, box))

Visual Example

📚 Documentation

• Model Training: The model was trained using a combination of two datasets: modanet and fashionpedia.
• Labels: The labels used by the model are ['bag', 'bottom', 'dress', 'hat', 'shoes', 'outer', 'top'].
• Performance: In the 96th epoch out of a total of 100 epochs, the best score was achieved with mAP 0.697400.

📄 License

This project is licensed under the MIT license.