TikTok has sparked a revolution of adult content, but with their crackdown on adult content, a number of TikTok porn spinoffs have now appeared. One of these apps, SikSok, is making waves for its use of parallel processing and AI-powered facial recognition to enhance your viewing experience.
The UI itself is fairly simplistic, consisting of a simple list view where users can browse through videos with a TikTok style interface. Some of the elements are custom made, including the search bar and the filters for searching, while other parts were pulled from the open source community.
The SikSok Backend
While SikSok.app is written with JavaScript, the backend is written in GoLang, and uses Tensorflow for its AI models. The app consists of three major components: the viewer (UI), the search engine (content server), and the recommendation engine (AI). The AI is used to identify what type of content and the gender of the person being filmed. This helps further improve the user’s experience by providing them with recommended videos that best match their preferences. What sets this apart from other TikTok porn apps, is how it uses its own custom backend for the content filtering algorithm and facial recognition.
Parallel Processing
SikSok also leverages distributed computing for improving performance. To do so, they first scale the workload across multiple servers and then distribute the results to all hosts. They utilize a system called sharding, which splits the database into shards and replicates data between the shards.
This allows them to run multiple instances of each service in parallel. By using multiple instances of the same service, each instance can be optimized independently, which improves overall performance. For example, say you wanted to process 10,000 photos per second. If you only had one instance, it would take the full 10,000 seconds to complete the task. However, if you split those tasks over five instances, it will now take only 2,500 seconds or 25 minutes to finish. In addition, each instance has the ability to perform faster than the original, due to not sharing resources between different instances.
This technique is commonly known as asynchronous execution, and is very popular in cloud computing. It’s typically used when dealing with large amounts of data that need to be processed simultaneously by many machines. While distributed computing is nothing new, SikSok’s implementation of it is unique because they’re able to scale out their service without having to deploy additional hardware. All they have to do is add more instances of the application.
Another benefit of using distributed computing is its ability to handle failure cases, such as server overloads. If an instance fails, the remaining instances continue processing, allowing the overall workload to keep up with demand. In the case of SikSok, if their content servers were ever overloaded, the recommendation engine could still process requests, since there’s no dependency on the server.
Facial Recognition Algorithm
SikSok’s facial recognition algorithm is able to detect the gender of people in the videos, as well as the age, using deep learning techniques such as recurrent neural networks (RNNs) and convolutional neural networks (CNNs). These algorithms are trained on thousands of samples, which allow them to achieve over 90% accuracy.
Another aspect of SikSok’s facial recognition algorithm is its ability to predict the popularity of a video. To do so, it uses a set of parameters that it measures based on the number of likes and views. Popularity prediction is an important use case for TikTok videos, which are often uploaded by celebrities. Since most of these videos get millions of views, predicting their popularity allows SikSok to recommend similar content to users in order to provide them with better recommendations.
Conclusion
SikSok is just one of many TikTok porn apps that have emerged in response to the restrictions placed on TikTok. By providing their users with improved performance and easy access to free and legal content, SikSok is poised to become one of the biggest TikTok porn apps in the near future.