RealChain Project Progress Update
2018/1/15–2018/1/21
Deal community members:
RealChain has been online for a week now; we will release weekly report from now on, including the latest status of RealChain development. Let’s take a look at the latest developments for the first week!
Completed work:
[Being listed on Exchange platform OKEX.com]
At 16:00 on 1/8/2018, RCT open for depositing.
At 16:00 on 1/9/2018, RCT begun to trade on platforms.
On 1/10/2018, withdrawing RCT became available.
[Working in progress]
Perform technical testing with more overseas’ exchange platforms.
Themis Module
Themis module is the first project of RealChain using Smart Contract”, which records the original information of commodity goods and marks the creator institutions.
[Completed work]
● Adjusted the recording format of commodity goods: adding original recording date time.
● Adjusted the creator institution format: adding “transcription” in institution types.
● Completed the fourth version of Smart Contract using singleton-contract module. Continued the comprehensive evaluation of the cost of GAS and optimization. Compared with previous version, the new version established 3% decrease in cost.
● Completed the second version of Smart Contract using multi-contract-list module. Continued the comprehensive evaluation of the cost of GAS and optimization. Compared with previous version, the new version solved the contract security breach issue caused by updating the institution.
● Testing case improvements: added four new testing cases, covering the test on new portal for updating the institution information
[Working in progress]
CI system improvements: planning to build more realistic testing environment for RCT data from Ethereum network.
Forseti Module
TaoDangPu currently has records over 300 thousand commodity goods. For historic reasons, this data was recorded with different meta-data and saved as different media types. It would cost prohibitive to manually input into the data into system. The Forseti module will solve this issue by connecting the traditional IT system to the Blockchain Infrastructure..
[Completed work]
● Conducted the data cleaning on the historical data using Chinese words segmentation algorithm.
● Filtered the previous scanned images using image recognition algorithm.
● Completed the alignment of the commodity goods basic information from video.
● Fixed the bug in the file extension convention. Successfully recognize the PNG images with .jpg suffix.
● Manually identified and filed 3,217 abnormal items.
[Working in progress]
Continue implementing the clustering algorithm based on word segmentation, automatically format the abnormal items.
Minerva module:
Only the metadata and SHA256 information of commodity information was saved on the blockchian of RealChain. In traditional IT system, real media data (regardless of being encrypted or not) is temporarily saved on media library system.
[Completed work]
● Added the additional security mechanism, encrypted data is no longer saved in block cache.
● Fixed the bug caused by HASH collision, the system will not restart infinitely because of that.
● Fixed the bug that caused the loss of metadata due to the failure of file creation attempt.
● Fixed the bug that caused the loss of data due to the power loss during the system’s power-up.
● Optimized the speed of cold boot.
[Working in progress]
Continue evaluating the CEPH distributed file system. Plan to complete the performance test on the small files.
Continue improving the test cases, current status 45/79.
Smart hardware module:
- Artificial Intelligence hardware development.
AI hardware is mainly used for image recognition, appraising rating and referencing. It greatly lowers the human error caused by the difference in lighting and scenes.
Since the smart hardware is not open source, it is still in the confidential research and development phase.
Current items publicly available:
● Sample machine for front collection is ready.
● The cost of the equipment has reduced to the civilian level. (the exact amount is confidential at this moment and will be released during the pre-sale session)
● Submitted the product design version 2 to the manufacturer.
- AI system, image recognition, material analysis, core comparison
● Version 1 is in the testing phase. Currently, it achieves over 90% accuracy rate. Successfully recognizes over 98% of over 40 handbag styles including luxury bands, LV and PRADA. Accuracy in watch recognition achieves higher rate, especially in high-end luxury brands, including Cartier, Rolex, etc. The testing recognition rate is over 98%.
● The future of the deep learning in AI system requires more sample data. R&D team is actively looking for more training data to improve the recognition rate. Hopefully, the accuracy will be 98% in version 2.0.
Note: Due to the patent and intellectual property protection requirements for the development of the smart hardware and AI system, this part of the research and development is closed source. Therefore, the content will not be updated until there is a breakthrough in the future.
See you next week!
RealChain Foundation Official Team
