Faceter is a system that makes video surveillance intelligent by providing computer vision, facial recognition, object detection and video content analysis to cameras in real time. In the next stage, all these abilities will be combined into one – “understand” the current situation and then react to it. The team calls this concept the "event security model."

Project Vision

After Faceter was adopted by the market, it could be used to maintain whitelists in various services; for example, on websites, employees are hired to do some housework. In addition to comments, the Faceter system can store links to blockchain records containing biometric data. The same method can be used to store records of shop visitors and to check the integrity of the buyer, seller, or messenger.
If all cameras connected to Faceter are integrated with the police and rescue services database, information about wanted criminals, missing children or people in difficulty can be immediately sent to their relatives and places of residence through the system. In such an emergency, police and special services will also be able to access the system to find "photos" of people in Faceter user's video profile, while not leaking confidential information, as the system will only use the camera's data and the owner agrees.

Project Features

1. Object and facial recognition and the resulting behavioral analysis greatly enhances the ability of video surveillance systems to detect potential customers and specific employees. The algorithms of this technology have been proven by LFW and MegaFace testing that they are the most trustworthy in the industry.
2. The convenient application of any computer or mobile device allows users to access the video surveillance system from anywhere without mastering complex and slow interfaces.
3. Fog Computing - performs identification calculations by allowing miners to enter a decentralized computing network, thus providing miners with higher revenue generation opportunities, which greatly reduces the cost of the product.
4. Blockchain-Faceter uses all opportunities for smart contracts to provide flexible and transparent payment options and identification mechanisms for fog computing networks. This is Faceter's contribution to the development of the crypto industry.
5. Machine Learning - Using neural networks, you can teach Faceter to react to changes in certain events.
6. Data Protection - Faceter will not expose the source video stream to a trusted environment; only anonymous data can be transferred to a decentralized network.
7. Faceter token is the "fuel" of the decentralized network, providing all participants with a flexible, transparent, cross-border solution mechanism.

Application Fields

B2C
The software is designed for families to use in apartments, office buildings, country houses, as well as in families’ shops and restaurants. Payment methods are based on monthly or annual subscriptions, and the rate of retention depends on the number of cameras used and the workload (the number of faces identified per unit of time). Faceter's solution to individual users:
– Improve safety level;
– Bearable;
– Widespread applicability (combined with smart home technology).
B2B
The software can be used in factories, warehouses, large enterprises, hotels, casinos, shopping malls, banks and catering chains, thanks to technical cooperation agreements with suppliers of home, office and industrial equipment, as well as smart device sensors and other components manufacturers management systems.
B2G
Municipal software can be used to analyze video streams on streets and roads, as well as in crowded places such as stations and airports. It can also be monitored in conjunction with the database.
Faceter assumes that the offender's face can be identified and run throughout the archive records database to reveal where the person has visited recently. The solution can automatically control compliance with regulations and regulations in transportation and public places, which will increase the speed of case solving and ensure higher-level automatic control of the police.

Technical Overview

1.Data processing
The Faceter software complex is designed in the form of a set of modules, each module is executed at a different system level and is responsible for each stage of platform operation. The language for ML component development is C++11. In addition, GPU acceleration is also actively used.
The phase of facial recognition processing
Receive and process decoded video streams from the camera;
Face detection (if the camera is not equipped with built-in face detection technology);
Facial frames are better recognized;
Use deep convolutional neural network to obtain the feature vector for everyone;
Vector clustering, comparison of eigenvectors with available databases and their location in the storage system;
Output through the API results.
2.Faceter distributed network
Due to decentralized services, which include rapid growth in blockchain-based businesses, Faceter plans to build its own computing networks, such as SONM or Golem, by attracting individual owners of GPU resources and using affordable solutions, rather than expensive cloud services. Therefore, there is no need to use expensive monopoly services such as Amazon, Microsoft, and Google.
The use of miner resources or other idle computing power is very beneficial for both those resources and consumers' owners. Even a rough comparison of the cost of renting a GPU server on Amazon Web Services and the amount of compensation paid to miners for the mining area shows a huge difference. Furthermore, the complexity of the network is adjusted every few months to reduce miners’ compensation and, after adopting the new consensus mechanism, no GPU computing is required at all.
The main participants in the Faceter network are the owners of GPU resources (nodes). The smart contract will be displayed with each node at the end. This contract will be used to provide compensation for computing power.
A decentralized environment requires special verification of the trustworthiness of the connected nodes. To ensure the quality of the calculations performed, the nodes with lower productivity will verify the performance of the other nodes and perform the calculations repeatedly. High-performance nodes should be fully loaded with tasks, while less powerful nodes will be given fragments of the same task.
The calculation results will be compared by smart contracts, and if participants have sufficient confirmation information, each of them will receive some of their rewards from the formed balance.
If the node responsible for verification receives a different compute hash, this will confirm the miners' dishonest behavior and will cause them to disconnect between new computers and redistribute accumulated rewards. Faceter calls this concept "verification identification".
The performance of tasks and the distribution of verification will be passed through special coordinator nodes (video hubs), which can also act as obfuscators for video hubs and sensitive data. These nodes will be located in a trusted area, unlike all other nodes that use worthless non-personal data.
3.Personal Data Protection
Users of video surveillance systems are usually the ones who connect the camera to the cloud platform. There are dozens of companies (sometimes millions) on the market that provide such cloud application services: Ivideon, Camcloud, XMEye, vMEyeSuper, etc.
All of these services provide CCTV cameras, but none of them actually develop clear policies for processing user data; often, there is no proper means to control access to the data.
Faceter's modular architecture allows users to work with the original video stream in a trusted environment: if possible, on the camera owner's device, on the Faceter's server, or in the camera itself, only obfuscated (non-personal) data is used to perform tasks on a decentralized network. In other words, these data cannot be converted to the original format by reference to the owner.
4.Faceter token usage model
The effective operation of a distributed network requires flexible, transparent and cross-border settlement mechanisms for all participants. The traditional methods of using fiat currencies are not feasible in this particular case, as they do not fully comply with any of the above requirements.
This means that every potential network participant must pass the extended KYC check, open their own bank to open a Forex account, and perform international transactions that comply with all the requirements of local law, making it nearly impossible to create a large-scale decentralized computing infrastructure.
The solution to this problem is Faceter's own token as the basis for making payments between smart video surveillance services and consumers of network participants. Integration plan to introduce the Faceter platform in blockchain technology is implemented in the following stages: Phase 1: Issuing tokens at Ethereum for free sale and internal economic use of the project, as well as paying the node owner (“miner”).
Phase 2: Development and rollout of blockchain and possible token migration. This step should be to solve the problem of distributed task allocation, improve the system's operation speed, ensure the independence of the Ethernet blockchain, and reduce the transaction costs in the platform.
System Function Model
Faceter's services can be paid with tokens or other payment tools (bank cards, e-wallets, cryptocurrencies). In order to ensure the normal operation of the latter, Faceter will implement a temporary conversion mechanism for payment tokens. Camera repair costs are set by Faceter in USD and can be revised quarterly.

Related links:
https://www.qukuaiwang.com.cn/news/6777.html

*The above content is compiled by the official account of non-small accounts. If reprinted, please indicate the source.