.png)
KEYWORDS: #AI #AIfuture #ecommerce #fashion #datasets #maadaa
As AI enters into every industry, the development of AI has taken the Fashion and E-Commerce business to a completely new paradigm. And the coronavirus pandemic has speeded up this AI transformation process.
For retailers and merchants, AI helped to reduce operational costs, boost sales, increase more efficient processes and customer satisfaction and enable sustainability to become the norm.
For customers, AI will help to enjoy an easier, more seamless, more convenient, more personalized, and faster-shopping experience than ever before.
Photo by ThisisEngineering RAEng on Unsplash
In such circumstances, maadaa.ai has been working on AI-in-Fashion since 2015. We would like to share some of our experiences and observations in this article, some interesting topics like virtual fitting, fake detection, and personalization.
This article will cover the following topics:
1. How AI will affect the future of Fashion and E-Commerce
2. The collection of Fashion & E-Commerce datasets
3. Trends in Fashion Datasets
1. How AI will affect the future of Fashion and E-Commerce
The fashion world has changed noticeably in the last few years. During the pandemic, our social connections were virtual, and this setting influences our social lives continuously.
Therefore, AI in Fashion and E-Commerce industries has also developed much faster and more efficiently.
1.1 Digital-only fashion business
With the rise of new digital and immersive AI technologies, along with increasing needs from forward-minded people who believe that clothing and accessories don’t always have to exist physically, digital-only fashion brands and design companies become mainstream choices.
Simply explained, digital fashion brands and design companies are similar to the costume and props departments of large video game companies. The purpose is to help customers (players and shoppers) to enhance their virtual personal.
For instance, Auroborosm, a luxury fashion house, is founded in 2018 in London, UK. Their products mirror today’s scientific and technical breakthroughs through physical couture and digital clothing.
Auroboros presents Biomimicry Digital Collection at London Fashion Week In Partnership with IODF
There are more digital fashion houses that make unique 3D apparel and virtual fashion accessories, such as The Fabricant, Republiqe, Tribute Brand, etc.
Not to mention many brands have already successfully held fashion runway shows in the Metaverse.
“There is a huge community of people and platforms that really care about fashion [in the Metaverse],” says Mishi McDuff, the founder of Blueberry, a digital wearables company that has sold over 20 million units of virtual clothing. [1]
Here is a video clip revealing what a digital fashion show means.
Inside Gary James McQueen’s first digital fashion show | Spotlight | Unreal Engine
1.2 See Now, Buy Now
Post-pandemic, consumers prefer shopping online from home now.
Shoppers can instantly take action when they shop online. The “See Now, Buy Now” tools on social platforms like Instagram and Pinterest bridge the gap between inspiration and purchase.
It also uses your browsing history and your purchase history to recommend more products that you’ll like.
This is already happening in our real world.
As the chief brand officer at Tommy Hilfiger, Avery Baker says, “It’s about delivering on the instant gratification that consumers are really seeking.”
However, outfit sizes are always a big obstacle in this case. Thanks to the rapid development of AI technologies, this problem is gradually being solved.
In the field of virtual fitting, companies like Virtusize enables online shoppers to buy the right size, either by measuring the clothes in their closet or by comparing specific brands and styles to their own.
source from virtusize official website
True Fit, as another example, partners with retailers to facilitate capabilities like AI-powered fashion discovery and exact-fit clothing and shoe recommendations.
Imaging this, while watching Netflix or playing video games, you will be able to freeze the frame, tap on one character’s outfit and it will jump to the purchase page immediately. Then you can buy the items you like in your size easily.
We believe the possibility will become reality soon, maybe just in a few blinks.
1.3 AI-powered Personalization
Once AI is in motion, forecasts of future demands become more precise.
With the help of more quantity, higher quality data, and better algorithms, AI-powered technologies will become trend influencers — predicting and designing style trends much quicker, more personalized and more automatically.
In April 2019, an AI “designer” called DeepVogue placed second overall and won the People’s Choice Award at China’s International Fashion Design Innovation competition.
An AI “Designer” Just Won Runner-Up in a Major Fashion Design Competition – RADII
China’s champ is back on top! UFC women’s strawweight fighter Zhang Weili reclaimed the division title yesterday…
radii.co
Today’s “AI Designer” is already helping brands create and iterate their designs more products quickly.
In this case, AI can help brands and retailers make smarter strategic decisions around product development and new business lines to bring the trend to market as quickly as possible.
Furthermore, AI and advanced technologies may also help to play in the push for sustainability in fashion and E-Commerce.
It allows brands and retailers to easily control their supply chain and help them know more about customers and their behaviors.
In this case, brands and retailers will know exactly how many items to produce and reduce the waste of manufacturing clothes that will never be worn
Once, by eliminating overproduction in the process, the energy and money required for storage and transport will have also declined with the surplus under control which means saving more time and cash.
With more rigorous, human-led training, AI assistance programs will continue to advance and become more accurate.
AI is for today, not just tomorrow.
2. The collection of Fashion & E-Commerce datasets
Based on our comprehensive accumulations of Fashion and E-Commerce technologies and application scenarios, maadaa.ai has been developing a series of standard datasets, which can help industrial and academic customers accelerate AI innovations in Fashion and E-Commerce.
These datasets include tasks such as clothing classification, clothing pattern classification, clothing fabric classification, clothing key point detection, clothing, human body semantic segmentation, scarf fabric segmentation, E-Commerce datasets, etc.
Fashion & E-Commerce Datasets Collection – maadaa.ai
3. Trends In Fashion Datasets
With the rapid development of e-commerce platforms, the clothing category has become a trillion-level market. With more than $30 billion in apparel sales on Amazon’s platform in 2018, it has surpassed Walmart to become the no. 1 apparel retailer in the United States. The COVID-19 epidemic in 2020 did not change the trend of market development, but further accelerated the pace of business reform and promoted the emergence of more new consumption patterns in the clothing market. Consumers are more receptive to online retail channels than before the epidemic, the digital transformation of physical retailers is in full swing, and new consumer innovations such as live delivery, online experience and the sharing economy are in full swing.
The demand characteristics of the clothing category are closely related to the gender, age, ethnicity, region, and consumption ability of the crowd. Buyers’ demand for clothing is actually diversified and fragmented. With the increase in clothing styles and the enrichment of application scenarios, the existing public datasets are evolving toward the following:
- Multiple Tasks.
- Larger Volume. To some extent, deep learning is based on data-driven approaches. Massive data plays an important role in model learning and understanding.
- Fine-Grained Annotation. It is manifested in the refinement of segmentation and classification. In the segmentation task, some accessories and other small objects are added to the mark, and the classification task expands the categories.
3.1 Multiple Tasks
Early Fashion datasets may only contain classification tasks. For example, fashion-Minst3 resampling JPEG images with a resolution of 762*1000 to obtain grayscale images with a resolution of only 28*28, such early datasets only marked the categories of clothing and other information suitable for classification tasks. After the launch of Deep Fashion, it added annotation information such as text description, Bbox and landmark, which also means that this dataset can be used for text retrieval and object detection task training. Md-fashion and Deep Fashion2 further provide Mask annotation and continue to expand the task category to semantic segmentation to complete more fine-grained object detection tasks. The emergence of Deep Fashion 3D adds 3D scanning data to the dataset of the Fashion series, which can deal with the 3D reconstruction tasks derived from the increasingly popular metasexes and virtual image scenes.
At present, the Fashion dataset already includes classification, segmentation, detection, generation, cross-modal retrieval, 3D reconstruction and other existing task annotations, and it may continue to increase in the future.
3.2 Larger Volume
In the task definition of supervised learning, deep learning can be regarded as a data-driven process, and massive and diverse data is of great significance to the improvement of network performance. With the development of time and the deepening of the fashion research, the number and volume of datasets are also increasing exponentially, not only because of the increasing demand for application scenarios but also because of the development of distributed training and the improvement of GPU storage capacity. The original fashion-Minst consisted of 6W training images and 1W test images, followed by Deep Fashion with 80W images and WAB with 104W images. Md-fashion has maintained the largest number of images at 360W so far.
Since the Fashion dataset is different from other datasets such as animal species and fruit species, the Fashion dataset has strong timeliness to a certain extent. For example, some latest styles, some new clothing materials, clothing accessories and so on will be updated over time, so the number of Fashion datasets will continue to be updated and tend to be larger in the future.
3.3 Fine-Grained Annotation
With the continuous development of technology and the continuous refinement and specialization of fashion application scenarios, datasets are also developing toward such a trend. For example, in the field of classification, there may be fewer than five categories before, but now MD-Fashion has increased to 80 categories. The previous classification was more focused on the attributes of clothing, such as short sleeves, jackets, and trousers, but now the dataset has added a more refined classification of clothing materials, such as cotton and silk. In the field of segmentation, the early DeepFashion series only had three types of segmentation, but now MD-Fashion supports up to 30 categories of classification labeling in one picture, which is more professional. In the aspect of detection, coordinate estimation is no longer limited to the object itself, and the key point estimation of clothing patterns is added.
The continuous specialization and refinement of the Fashion application field require more fine-grained labeling information to better meet downstream tasks.
Reference List:
[1] https://www.refinery29.com/en-gb/2022/02/10884293/virtual-fashion-shows-metaverse
Further reading:
AI in fashion & E-Commerce Industry: application scenarios and technologies
AI Datasets for Fashion & E-Commerce: Open vs. Commercial and the Trends
AI for virtual fitting: inspired by datasets （Open & Commercial）
AI for fake Detection in Fashion and E-commerce industries: The related open & commercial datasets
AI-powered personalization of E-commerce and Fashion: open and commercial datasets
Face Parsing: use cases and open datasets
