Inspiration
The idea for our feature comes from the clothing recognition capability found on platforms like Taobao. We want to bring this innovative feature to TikTok, allowing users to easily identify clothes in videos with just a click. Additionally, we are inspired by the convenience this technology offers in bridging the gap between viewing and purchasing, enhancing user engagement.
What it does
Our AI recognition feature enables users to discover clothes within TikTok videos through a click. The AI Smart Recommendation System accurately recognizes the clothes, lists them out, and directs users to the TikTok shop to purchase the items. The system also provides alternative suggestions based on the user's style preferences and previous interactions.
How we built it
We trained our model using a labeled dataset. This involves showing the AI various images (e.g., a shirt) with corresponding labels (‘shirt’) to teach it to identify clothing correctly. Our model also underwent rigorous testing with a separate dataset to ensure high accuracy. We utilized data augmentation techniques to improve the model's robustness and performance across different lighting conditions and angles.
Challenges we ran into
One major challenge was our initial lack of knowledge in AI training. We had to learn everything from scratch, from feature extraction to model tuning and data augmentation. Additionally, obtaining a diverse and high-quality dataset was challenging, requiring significant time and resources for proper labeling and cleaning.
Accomplishments that we're proud of
We successfully built a model that accurately recognizes and lists clothes from TikTok videos, enhancing the shopping experience for users. We also integrated a user-friendly interface that seamlessly blends with TikTok's existing features, making the technology accessible to all users.
What we learned
We learned the importance of a well-labeled training dataset and the impact of data augmentation on improving model performance. Cleaning and refining the dataset were crucial steps in enhancing accuracy. We also gained insights into user behavior and preferences, which informed the development of our smart recommendation system.
What's next for AI Fashion
We plan to further personalize the shopping experience by incorporating smart recommendation systems. These systems will use data from visual searches to suggest similar and personalized fashion choices for users. Additionally, we aim to expand our feature to include accessories and other fashion items, creating a comprehensive shopping experience within TikTok.
Log in or sign up for Devpost to join the conversation.