This project harnesses the power of trained fashion models to detect objects within Pinterest images, meticulously correcting detection errors and optimizing detection accuracy by leveraging multiple models and mixed methodologies. The result is a finely tuned dataset aimed at enhancing fashion object detection across five principal categories.
provided dataset is in Yolo format, and the example models provided is trained on yolov8n with 100 epoch, and the image size is 640
- Object Detection: Utilize state-of-the-art pre-trained fashion models to identify fashion items within Pinterest images.
- Error Correction: Implement a rigorous process to correct any detection mistakes, ensuring the dataset's high accuracy and reliability.
- Datasetion: Employ a combination of multiple pre-trained models and diverse techniques to refine and enhance object detection performance.
- Categorized Dataset: The Dataset is meticulously organized into 6 main categories, catering to a wide range of fashion object detection applications.
- Tops
- Bottoms
- Dresses
- Outerwear
- Shoes
- Accessories