Future Work:

One of the authors of the first model, X. Yang, stated that users present in the experiment “expressed the desire for more detailed colors,” and pattern categories, such as rose red instead of simply red. Some blind participants also expressed the desire to have the camera on a cap instead of sunglasses, due to the extra weight of the web cam. They also expressed the desire to have this function available on mobile phones instead for efficiency, since most people carry mobile phones on their person.

The authors of the second model mentioned that their system can have applications for online shopping, and that “a lot of companies are starting to utilize these technologies which help them get more insight into their customer base while providing more opportunities to those with visual impairments.”

Conclusion:

In conclusion, the topic of computer vision for fashion matching is still fairly unexplored, particularly when applied to assisting the visually impaired. Three different models were observed, each using different techniques and implementations for various fashion matching purposes, including clothing type classification, color recognition, and pattern and attribute recognition. If we take elements from some of these models, it can be possible to implement an application that fulfills the goal to assist the visually impaired in outfit-making decisions. For instance, image capturing and voice commands from the first method can be implemented on a mobile device, while clothing type and multi-attribute recognition can be taken from the second model in order to achieve the most descriptive and meaningful results.