AI for fake Detection in Fashion and E-commerce industries: Use Cases and technologies
#FakeDetection #Fashion&eCommerce #openDatasets #AI #Counterfeits
The combination of Artificial intelligence and fashion has become a very active research topic in recent years.
According to consultancy Bain, sales of luxury goods are set to rise at least 5% this year as shoppers in the United States and Europe continue to snap up high-end watches, jewelry and shoes despite the conflict in Ukraine and soaring inflation.[1]
Unfortunately, The market for counterfeit goods is also rising.
Counterfeits are quite painful for both luxury and famous fashion brands but they can be even more of a headache for digital re-sellers.
Since AI applies to every industry in the world. So in fashion and e-Commerce industries, how does AI work for a specific topic: fake detection (or counterfeits recognition)? In this article, we will talk about:
1. Examples of how AI technologies recognize counterfeits.
2. The essential technologies behind fake detection.
3. The related open & commercial datasets
Examples of how AI technologies recognize counterfeits
E-commerce sites and second-hand markets are the hardest hit by the sale of counterfeit goods. Therefore, It is especially important to use advanced and smart detection technologies to screen for fakes.
Fortunately, machine learning and deep learning methods can find their use in anti-counterfeiting applications.
Here are some state-of-the-art examples which are working on finding innovative solutions to curb the sale of counterfeit products.
Amazon is tackling the issue by putting in place machine learning and brand registry to weed out fake goods and sellers.
In this case, brands with registered trademarks can apply to the Brand Registry Program. It helps sellers identify and report infringements of intellectual property rights.
Goat and Entrupy, these two companies have developed anti-counterfeit algorithms based on immense databases of information on top luxury brands. These datasets are used to look for the tiniest of inconsistencies to protect the rights of both brands and customers.
For instance, Entrupy, with a scanner and an application, can instantly detect imitation designer bags by taking microscopic pictures that take into account details of the material, processing, workmanship, serial number, and wear/tear.
In order to create its enormous database, they first collected bags they believed to be authentic, dating as far back as 80 years.
Chinese Retail Giant Alibaba set up the Alibaba Anti Counterfeiting Alliance (AACA) in 2017 to tackle the counterfeiting issue.
Cooperating with notable companies like Cisco, Alibaba is working hard to raise awareness of the dangers of counterfeiting and hold counterfeiters accountable for their illegal practices.[2]
Not only famous and luxury brands need AI technology to recognize counterfeits.
Hundreds and thousands of smaller e-commerce marketplaces would benefit by engaging expert AI development services to help them to detect fake goods and thus ensure buyer trust as well as their reputation.
So, will 2023 be your year to meet such a huge demand for counterfeits recognition?
The enable technologies behind fake detection
Artificial intelligence companies use computer vision to recognize fakes. Data scientists design machine learning algorithms to detect details. These details separate legit items from counterfeits.
Therefore, fake detection includes a variety of applied technologies, which can be summarized as follows:
1. Image Classification
Through the image classification technology, the clothing such as: skirts, short sleeves, hats, shorts and other categories are automatically classified. The corresponding application scenarios of image classification are: automatic clothing sorting, content recommendation attribute classification, etc. Classification technology can also be divided into: classification based on clothing style, classification based on clothing attributes.
Classification is based on clothing styles, often targeting categories for specific clothing such as shoes, tops, pants, etc. The pattern of feature extraction network, fully-connected network, and multi-classification loss is often used to recognize clothing classification. Feature extraction networks include the VGG series, Resnet series, and Inception series. The Loss function uses Focal Loss and multi-class cross entropy Loss, depending on how balanced the dataset is.
2. Image Segmentation
For a fashion picture, the image segmentation technology can distinguish the clothes, pants, and various accessories of the figure in the picture at the pixel level.
Mask R-CNN[3] is often used as the Baseline network structure for image semantic segmentation task. Mask R-CNN can be decomposed into Resnet-FPN, RPN, Faster RCNN ,and Mask branch. Resnet-FPN performs convolution operations on the input images and extracts features, and then obtains feature images of different scales through the FPN layer. RPN selects the most suitable feature map for the candidate region. Finally, the network is divided into two branches, one is the traditional Faster RCNN detection and classification branch; In addition, the Mask prediction task is completed for the unique semantic segmentation branch of the Mask network.
The image segmentation technology is mainly evaluated by MPA, MIoU ,and so on. MPA is called Mean Pixel Accuracy, which mainly measures the Accuracy of predicted semantic segmentation map and Ground Truth on pixels. MIoU becomes mean IoU, that is, the average intersection ratio, that is, the intersection of predicted region and actual region is divided by the union of the predicted region and actual region. In this way, the IoU of a single category can be calculated. Then repeat this algorithm to calculate IoU of other categories, and then calculate their average. MIoU better reflects the accuracy of predictions between regions.
3. Detection
Detection can be divided into target detection and key point detection, and both of them are coordinate points that need to regression clothing according to the picture. It has great technical value in clothing search and recognition.
Yolo[4] is a typical algorithm for target detection. The input first divides an image into a grid, and if the center of an object falls in the grid, the grid is responsible for predicting the object. Each network needs to predict the location information and confidence information of a BBox, and one BBox corresponds to four location information and one confidence information. Confidence represents the confidence of the predicted box containing object and how accurate the box is. The target window can be predicted according to the previous step, and then the target window with low possibility can be removed according to the threshold value. Finally, the NMS can remove the redundant window.
Key point detection techniques can be divided into regression faction and heat map faction. The key point detection of regression faction processing is generally realized by extracting feature network, connecting the whole connection layer and predicting the coordinates of key points. Loss functions are usually MSE or MSE variants. Each channel in the diagram represents a key point of a category. If there are several key points of several categories, there are several channels. The key point position on a channel map is a two-dimensional Gaussian distribution centered around it, and the pixel value of the remaining positions is 0. The result of network prediction is also a thermal diagram. Generally, the most direct way to extract coordinates is to extract the point in a channel whose pixel response is greater than a certain threshold and has the largest response, and the coordinate of this point is the coordinate of the key point of this category.
The related open & commercial datasets
There is an old Chinese saying that missing an inch, a thousand miles away.
This statement is especially suitable for the field of artificial intelligence. In order to train AI models effectively, the quantity and quality of data are the key issues.
Tiny differences in data accuracy will directly affect the accuracy of AI models which might become massive disasters.
maadaa.ai has carefully collected datasets for fake detection or let’s say counterfeits recognition, both open and commercial. Please click the image below for more Info:
maadaa.ai Open Datasets:
1. Unleash your creativity with this AI Photo & Video Editing Open Dataset. Explore the power of AI in transforming visual content!
https://maadaa.ai/datasets/OpenDatasetDetail/AI-Photo-Video-Editing-Open-Dataset
2. Fuel your research and analysis with a diverse range of open datasets. Discover a world of data at your fingertips!
https://maadaa.ai/datasets/OpenDataset
3. Dive into the world of fashion and e-commerce with this open dataset. Gain insights into the latest trends and revolutionize your strategies!
https://maadaa.ai/datasets/OpenDatasetDetail/Fashion-&-E-commerce-Open-Dataset
Reference List:
[1] Mimosa Spencer Luxury sales to grow at least 5% this year — Bain, 2022. https://www.reuters.com/business/retail-consumer/global-luxury-outlook-still-strong-sales-grow-least-5-this-year-consultancy-2022-06-21/
[2] Al Palladin, Cisco and Alibaba Collaborate against Counterfeiting, 2022. https://blogs.cisco.com/smallbusiness/cisco-and-alibaba-collaborate-against-counterfeiting
[3] He, Kaiming, et al. “Mask r-cnn.” Proceedings of the IEEE international conference on computer vision. 2017.
[4]Redmon, Joseph, et al. “You only look once: Unified, real-time object detection.” Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
[5]He, Kaiming, et al. “Deep residual learning for image recognition.” Proceedings of the IEEE conference on computer vision and pattern recognition. 2016.
