The DeepFashion dataset is a large-scale clothes database, which has several appealing features: Clothing Category and Attribute Prediction, In-shop Clothes Retrieval Benchmark, Consumer-to-Shop Clothes Retrieval Benchmark, and Fashion Landmark Detection Benchmark, collected by the Multimedia Lab at the Chinese University of Hong Kong. However, for our project, we’ll use only the Category and Attributes Prediction dataset because we’re going to work on detecting and classifying clothing in existing images, and even generating new similar images. To follow along, download the dataset. Category and Attributes Prediction is a huge dataset that contains images of clothes segregated into highly specific categories by different attributes. For example, blouses with sleeves are considered different from sleeveless ones. For this project, we made our own data subset, reducing the volume of images and category specificity, for simplicity and lower computation costs. We reduced our classification from DeepFashion’s original 46 categories to 15 categories. Then, we selected 500-700 images from each of our simplified categories.
Link for the whole Sourcecode download: https://www.codeproject.com/KB/AI/5297227/DeepFashion.zip
Links for articles:
- https://www.codeproject.com/Articles/5297227/Deep-Learning-for-Fashion-Classification
- https://www.codeproject.com/Articles/5297322/Preparing-Data-for-AI-Fashion-Classification
- https://www.codeproject.com/Articles/5297327/Fine-tuning-VGG16-to-Classify-Clothing
- https://www.codeproject.com/Articles/5297329/Running-AI-Fashion-Classification-on-Real-Data Generate New Fashion Designs with a Generative Adversarial Network (GAN)
- https://www.codeproject.com/Articles/5297333/Training-and-Running-a-GAN-for-Fashion-Design-Gene