Dataset Card for "deepfashion"

More Information needed

Deepfashion

## Overview

Deepfashion is a dataset for object detection tasks - it contains Clothes annotations for 240 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [Public Domain license](https://creativecommons.org/licenses/Public Domain).

Disclaimer: We do not own this dataset. DeepFashion dataset is a public dataset which can be accessed through its website. This dataset was used to evaluate Marqo-FashionCLIP and Marqo-FashionSigLIP - see details below.

  Marqo-FashionSigLIP Model Card

Marqo-FashionSigLIP leverages Generalised Contrastive Learning (GCL) which allows the model to be trained on not just text descriptions but also categories, style, colors, materials, keywords and fine-details to provide highly relevant… See the full description on the dataset page: https://huggingface.co/datasets/Marqo/deepfashion-multimodal.

DeepFashion2 Dataset

About this Dataset

DeepFashion2 is a comprehensive fashion dataset. It contains 491K diverse images of 13 popular clothing categories from both commercial shopping stores and consumers. It has a total of 801K clothing items, where each item in an image is labeled with scale, occlusion, zoom-in, viewpoint, category, style, bounding box, dense landmarks, and per-pixel masks. There are also 873K Commercial-Consumer clothes pairs.

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F22418539%2Fc38fe86b17de5c19d044f4ffb7d7c203%2Fdeepfashion2_bigbang.png?generation=1763420762376839&alt=media" alt="">

Tasks Enabled by this Dataset

The richness of the annotations makes DeepFashion2 suitable for multiple complex tasks, often simultaneously (as in a multi-head model like a Keypoint R-CNN):

1. Object Detection: Find the bounding_box for each clothing item.
2. Instance Segmentation: Generate a per-pixel segmentation mask for each item.
3. Landmark (Keypoint) Estimation: Localize the 294 potential landmarks for each item. This is the core task for virtual try-on and detailed clothing analysis.
4. Commercial-Consumer Retrieval: Given a "consumer" photo, find the matching "shop" item.

The dataset is split into a training set (391K images), a validation set (34k images), and a test set (67k images).

Figure 1: Examples of DeepFashion2. https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F22418539%2Fde20008e6aa8f794c4b30d8333df5b55%2Fannotation.jpg?generation=1763420915372772&alt=media" alt="">

From (1) to (4), each row represents clothes images with different variations. At each row, the left three columns represent clothes from commercial stores, while the right three columns are from customers. In each group, the three images indicate three levels of difficulty with respect to the corresponding variation. Each item is annotated with landmarks and masks.

Data Organization

The data is organized into train, validation, and test sets. Each image (.jpg) has a corresponding annotation file (.json) in the annos directory.

• Training images: train/image
• Training annotations: train/annos
• Validation images: validation/image
• Validation annotations: validation/annos
• Test images: test/image

Annotation JSON Structure

Each .json annotation file is organized as a dictionary with image-level information and a variable number of item entries.

• source: A string, where 'shop' indicates the image is from a commercial store and 'user' indicates the image is from a consumer.
• pair_id: A number. Images from the same shop and their corresponding consumer-taken images share the same pair_id.
• item_1, item_2, ... item_n: Each item found in the image has its own entry with the following keys:
  • category_name: A string indicating the category of the item.
  • category_id: A number (1-13) corresponding to the category name:
    1. short_sleeved_shirt
    2. long_sleeved_shirt
    3. short_sleeved_outwear
    4. long_sleeved_outwear
    5. vest
    6. sling
    7. shorts
    8. trousers
    9. skirt
    10. short_sleeved_dress
    11. long_sleeved_dress
    12. vest_dress
    13. sling_dress
  • style: A number (0, 1, 2, ...) to distinguish between clothing items from images with the same pair_id. See the "Understanding Pairs" section for more details.
  • bounding_box: [x1, y1, x2, y2] coordinates of the upper-left and lower-right corners.
  • landmarks: [x1, y1, v1, ..., xn, yn, vn], where v is visibility:
    • v=2: visible
    • v=1: occlusion
    • v=0: not labeled
  • segmentation: [[x1, y1, ... xn, yn], [poly2], ...] A list of polygons.
  • scale: 1 (small), 2 (modest), or 3 (large).
  • occlusion: 1 (slight/none), 2 (medium), or 3 (heavy).
  • zoom_in: 1 (no), 2 (medium), or 3 (large).
  • viewpoint: 1 (no wear), 2 (frontal), or 3 (side/back).

Landmark & Skeleton Definitions

A total of 294 landmarks are defined across the 13 categories. The numbers in the figure below represent the order of landmark annotations for each category.

Figure 2: Definitions of landmarks and skeletons. https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F22418539%2Ff3d67659bb9e66b17d6238ace90f1fc2%2Fcls.jpg?generation=1763420957293495&alt=media" alt="">

Understanding Pairs (for Retrieval)

A key feature of this dataset is the link between "shop" (commercial) and "user" (consumer) images for clothing retrieval.

• Images with the same pair_id are...

DESCRIPTION 📝

DeepFashion2 Original (with Dataframes) includes the following: - The original DeepFashion2 Dataset, comprising JSON files with boundary box annotations, segmentation masks, and landmark points. The /annos sub-folder contains these files, while the /images sub-folder holds images for training and validation. The test folder includes images without annotation information. - The dataframes that have been generated through the steps in the notebook Resizing DeepFashion2 (256x256) and Basic EDA, converting all the JSON information for every image into 3 dataframes - train, test and validation for the ease of pre-processing, visualisation, and modelling.

ORIGINAL SOURCE 👚👔

DeepFashion2 Dataset

CITATION 📖

@article{DeepFashion2,
 author = {Yuying Ge and Ruimao Zhang and Lingyun Wu and Xiaogang Wang and Xiaoou Tang and Ping Luo},
 title={A Versatile Benchmark for Detection, Pose Estimation, Segmentation and Re-Identification of Clothing Images},
 journal={CVPR},
 year={2019}
}

NOTE - THE CITATION INFORMATION SHOWN BELOW UNDER DOI Citation IS AUTOGENERATED BY KAGGLE. USE THE ABOVE BIBTEX WHILE CITING THE DATASOURCE

The DeepFashion-MultiModal Dataset is a leading dataset for fashion AI and computer vision, featuring high-quality, aligned images and textual attributes for fashion recognition, retrieval, and generative AI applications.

The dataset is the re-organized and re-labeled version of the In-shop Clothes Retrieval Benchmark of DeepFashion. It includes 13,752 pairs of images and masks.

The original data was presented in the form of a deep file hierarchy and had to be re-organized as only image and mask folders under the data directory. All masks had three channels, they were reduced to one channel. Not all images had masks in the original dataset. Images without masks were discarded. You can find the script that achieves these tasks here.

Liu, Ziwei and Luo, Ping and Qiu, Shi and Wang, Xiaogang and Tang, Xiaoou, DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2016.

Original source: DeepFashion: In-shop Clothes Retrieval

One can find the notebook where this dataset is used.

License info: http://mmlab.ie.cuhk.edu.hk/projects/DeepFashion/DeepFashionAgreement.pdf

DeepFashion MultiModal Parts2Whole

  Dataset Details





  Dataset Description

This human image dataset comprising about 41,500 reference-target pairs. Each pair in this dataset includes multiple reference images, which encompass human pose images (e.g., OpenPose, Human Parsing, DensePose), various aspects of human appearance (e.g., hair, face, clothes, shoes) with their short textual labels, and a target image featuring the same individual (ID) in the same outfit… See the full description on the dataset page: https://huggingface.co/datasets/huanngzh/DeepFashion-MultiModal-Parts2Whole.

DeepFashion Clothing

## Overview

DeepFashion Clothing is a dataset for classification tasks - it contains Clothing annotations for 3,169 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [MIT license](https://creativecommons.org/licenses/MIT).

Original data contains 44k. This one only contains images from front.

DeepFashion-MultiModal

Text2Human: Text-Driven Controllable Human Image Generation
Yuming Jiang, Shuai Yang, Haonan Qiu, Wayne Wu, Chen Change Loy and Ziwei Liu
In ACM Transactions on Graphics (Proceedings of SIGGRAPH), 2022.

From MMLab@NTU affliated with S-Lab, Nanyang Technological University and SenseTime Research.

https://github.com/yumingj/DeepFashion-MultiModal/raw/main/assets/logo.png">

[Project Page] | [Paper] | [Code] | [Demo Video]

DeepFashion-MultiModal is a large-scale high-quality human dataset with rich multi-modal annotations. It has the following properties: 1. It contains 44,096 high-resolution human images, including 12,701 full body human images. 2. For each full body images, we manually annotate the human parsing labels of 24 classes. 3. For each full body images, we manually annotate the keypoints. 4. We extract DensePose for each human image. 5. Each image is manually annotated with attributes for both clothes shapes and textures. 6. We provide a textual description for each image.

@article{jiang2022text2human,
 title={Text2Human: Text-Driven Controllable Human Image Generation},
 author={Jiang, Yuming and Yang, Shuai and Qiu, Haonan and Wu, Wayne and Loy, Chen Change and Liu, Ziwei},
 journal={ACM Transactions on Graphics (TOG)},
 volume={41},
 number={4},
 articleno={162},
 pages={1--11},
 year={2022},
 publisher={ACM New York, NY, USA},
 doi={10.1145/3528223.3530104},
}

@inproceedings{liuLQWTcvpr16DeepFashion,
 author = {Liu, Ziwei and Luo, Ping and Qiu, Shi and Wang, Xiaogang and Tang, Xiaoou},
 title = {DeepFashion: Powering Robust Clothes Recognition and Retrieval with Rich Annotations},
 booktitle = {Proceedings of IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
 month = {June},
 year = {2016}
 }

The above dataset is derived and pre-processded from the DeepFashion Multimodal Dataset for :

https://github.com/yumingj/DeepFashion-MultiModal Text2Human: Text-Driven Controllable Human Image Generation Yuming Jiang, Shuai Yang, Haonan Qiu, Wayne Wu, Chen Change Loy and Ziwei Liu In ACM Transactions on Graphics (Proceedings of SIGGRAPH), 2022.

The above dataset is specifically preprocessed based on Characterisitc Input for Image -Text Pairs consisting: - -Gender Classification - -Feature Engineering - - ViT Input Compose

Dataset Contains : - - fashion_model.pth (Trained on the given dataset for pairing) - - male_fashion (Image Directory for all Male Images) - - female_fashion (Image Directory for all Female Images) - - df_male.csv (Text for all Male Images) - - df_female.csv (Text for all Female Images) - - female_front.csv (Text specified for Front-full body Images) - - features.db (sqlite database with Image-Text pairs)

Dataset Card for "deepfashion-multimodal-descriptions"

More Information needed

zhengchong/DeepFashion dataset hosted on Hugging Face and contributed by the HF Datasets community

Deepfashion: Powering robust clothes recognition and retrieval with rich annotations.

ThanThoai9x/Deepfashion dataset hosted on Hugging Face and contributed by the HF Datasets community

Introducing the Fashion Apparel Image Classification Dataset for Convolutional Neural Networks (CNN), a carefully curated collection of clothing images specifically designed for CNN-based image classification tasks. This dataset features 5,413 high-quality images of various clothing items in two primary colors: black and blue. The images are categorized into 10 distinct classes:

• black_dress: 450
• black_pants: 871
• black_shirt: 715
• black_shoes: 766
• black_shorts: 328
• blue_dress: 502
• blue_pants: 798
• blue_shirt: 741
• blue_shoes: 523
• blue_shorts: 299

Each category contains a substantial number of images, ranging from 299 to 871, ensuring a well-balanced and diverse dataset for robust model training and testing. The dataset showcases a wide variety of clothing styles, designs, and textures, making it an ideal resource for developing and refining CNN models for fashion apparel image classification.

This Fashion Apparel Image Classification Dataset for CNN is perfect for researchers, developers, and students working on computer vision, image processing, and deep learning projects in the fashion and apparel domain. Use it to train and test your CNN models for object detection, image segmentation, and clothing classification tasks. Explore this dataset and elevate your fashion apparel image classification projects to new heights.

This dataset contains images of clothing items scraped from Carousell, an online marketplace, specifically curated for image classification tasks. It includes a diverse set of classes representing different types of clothing, making it an excellent resource for machine learning and computer vision projects. The dataset is organized into the following 15 classes: - Blazer - Celana_Panjang (Long Pants) - Celana_Pendek (Shorts) - Gaun (Dresses) - Hoodie - Jaket (Jacket) - Jaket_Denim (Denim Jacket) - Jaket_Olahraga (Sports Jacket) - Jeans - Kaos (T-shirt) - Kemeja (Shirt) - Mantel (Coat) - Polo - Rok (Skirt) - Sweter (Sweater)

The images in this dataset represent various styles, textures, and colors, offering a comprehensive resource for training models to recognize and classify clothing categories. It is ideal for tasks such as building fashion recommendation systems, creating virtual try-on applications, or studying visual trends in fashion e-commerce. Whether you are an enthusiast or a professional, this dataset can help explore and experiment with deep learning techniques in the realm of fashion.

1) Data Introduction • The Clothes Dataset is a classification dataset composed of clothing images collected from Carousell, an online marketplace. It consists of 15 clothing categories (e.g., T-shirts, shirts, jackets, dresses, etc.), and can be utilized in various computer vision tasks such as clothing classification, fashion recommendation systems, virtual try-on applications, and fashion trend analysis.

2) Data Utilization (1) Characteristics of the Clothes Dataset: • The dataset contains clothing images with a wide range of colors, textures, and styles, making it highly suitable for realistic fashion item recognition tasks.

(2) Applications of the Clothes Dataset: • Clothing image classification model development: Can be used to train deep learning models that automatically classify images into 15 clothing categories. • Fashion recommendation and virtual try-on systems: Useful for building AI models that recommend appropriate clothing based on user preferences or body silhouettes.

DaHaDaHa/deepfashion dataset hosted on Hugging Face and contributed by the HF Datasets community

Our fashion dataset is composed of information about 24,752 posts by 13,350 people on Instagram. The data collection was done over a month period in January, 2015. We searched for posts mentioning 48 internationally renowned fashion brand names as hashtag. Our data contain information about hashtags as well as image features based on deep learning (Convolutional Neural Network or CNN). The list of learned features include selfies, body snaps, marketing shots, non-fashion, faces, logo, etc. Please refer to our paper for the full description of how we built our deep learning model.