Client Overview

Our client, a leading beauty brand offers the largest color portfolio when it comes to lipstick shades, beauty experiences, and other products. The products come with scientifically-advanced formulas and trending shades & textures at affordable pricing. Their products are produced and packed in Germany, France, Korea, Italy & the United States of America, and are currently available at both, online and offline stores.

With the fear of an on-going pandemic all across the globe, it was obvious that customers would avoid physical stores for a while. Therefore, our client wanted their website to provide a ‘virtual try-on experience’ to their customers so that they could try the products virtually on their faces with an easy click of a button.

Saffron Tech helped the client build a website powered with Artificial Intelligence (AI) and Machine Learning (ML) which enabled them to augment decision making and automate the process of purchase. This, in turn, helped them smoothen their process of doing online business because their customers were now able to make confident decisions once they got to try all the products virtually.

Challenges/Aim

The sole purpose of this project was to provide a real-time ‘Virtual Product Try-On Technology’ to our clients. The aim of this project was to assist our client’s customers with their shopping experience and make it virtually wholesome. This helped boost the purchase intent of the customers and decrease the number of customers dropping out of their buying cycle. The technology helped their customers make a more confident purchase decision. Also, there would be no fear of contamination.

Solution Offered

Saffron Tech proposed a ‘BeautyAR’ which is a virtual lipstick try-on technology for their website. A POC was built for national cosmetics retailers to directly integrate with their website.

Technology Used

Python Language – Saffron Tech used ‘Python code’ as it is well-suited for A.I. because Python makes the construction of AI models much simpler. It brings a broad choice of systems and libraries; it has also been used by popular brands like LensKart. It can perform an array of complex AI errands and help in constructing AI models at a rapid speed.

Deep Learning Frameworks – Python exhibits a diverse arrangement of libraries for computer reasoning and Artificial Intelligence. Therefore, for deep framework learning, Saffron Tech used Pytorch, Tensorflow, and Keras to base the feature.

Numerical Computation – We used NumPy for boosting information examination and for establishing optimum logical registering. We also used Pandas for information examination. The combination of NumPy and Pandas enabled us to bring our clients the ultimate structure they required.

Machine Learning – For the Machine Learning pipeline, we used sci-kit-learn, a learning library for Python Language which offered various dynamic tools for machine learning such as regression, classification, clustering, and more.

Image Processing – We operated image processing with OpenCV or Open Source Computer Vision Library because it provides more than 2500+ optimized algorithms. Skimage for image pre-processing. We also used PIL, to provide image editing functionalities.

DevOps – We used DevOps with Docker because it makes load balancing simpler with pre-installed service concepts, AWS for Cloud Formation and service integration, and GCP or Google Cloud Platform.

Conclusion

Today, with the help of Artificial Intelligence and Machine Learning algorithms, we can develop trained systems that can detect certain features, objects, and people.

The biggest confusion for customers of eCommerce beauty products is whether the specific product will look good on them or not. With the help of the ‘upload photo’ button or the front camera, customers would be able to try the lipstick colors on their face virtually. Saffron Tech was able to successfully deliver the required technology to base this exceptional feature and help our client provide unparalleled shopping experience.