![enterx7](https://pic-cc.com/data/avatars/m/87/87506.jpg?1663946100"){kind=avatar}
- Joined
- 22.09.22
- Messages
- 3
- Reaction score
- 1
- Points
- 1
==========================================
Black Data Science, as well as ...
- development of logs, text arrays for CC/CVV;
- reading by neural network from screenshots of data on CC/CVV;
- Validation CC/CVV, transformation into the format of checks, shops;
- custom requests;
- other cooperation in data processing.
High speeds. Multi -flow programming.
Neural networks work on GPU.
==========================================
FAQ:
*** How does the scheme with the development of screenshots work, the search for CC/CVV in logs/text massifs?
1) Send a pack of screenshots/text_fails/logs on which there may be of interest to the data.
2) you get a signal that the development is completed and the screen with part of the finished results.
3) send funds to the guarantor.
4) Get a dump base (SQL) or CSV.
*** How does a circuit with cc/CVV data validation work, converting into the format of checks, shops?
In general, similarly with the previous point.
*** What is a share of suitability for working out screenshots?
Let there be 100 screenshots with a resolution of 1024 x 768 px. On 5 screenshots, CC/CVV data were found.
The share of fitness will be: 5/100 = 0.05
*** What is a percentage of fitness for working out text arrays / logs?
The percentage of fitness for text arrays / logs is calculated in a similar way to the previous point. In this case, the calculation basis is the number of CC/CVV detected by 1 megabyte of text data.
*** What is the minimum permissible fraction of the fitness?
Work out screenshots / text arrays / logs with zero fraction of suitability - so -so lesson ... for obvious reasons. At the same time, it is quite difficult to determine the percentage of fitness in advance. Only indirectly, through the origin of the data. Therefore, if you intend to give log/text arrays in the development, you must be sure that the source deserves attention. The minimum permissible fraction of fitness is taken equal to 0.03.
If you are not sure of the source, do not waste time.
*** It is necessary to validate the lines CC/CVV. Next, lead to shopping format. Does the share of suitability affect the cost of processing?
Does not affect. The cost is determined based on the number of lines transferred to processing.
*** How to get potentially interesting screenshots?
Typically, screenshots are extracted in two ways:
- using the styller;
- Using spiders robots that roam the fields of services such as Lightshot.
At the same time, the screenshots received in the second way have a lower percentage of fitness. Castamize Targets for spiders to increase the fraction of suitability.
A more complete list of graphic services, for example, here:
esputnik.com/blog/13-servisov-dlya-sozdaniya-skrinshotov
It is also recommended for reading:
vk.com/@haccking1-bluzhdaem-prostora-github-dyry-skripty-qody-i-chartov
*** How to write a spider for screenshots?
The starting point can be:
LightShot и чужие скриншоты
Сегодня я наткнулся на одну «уязвимость» сервиса для моментальных скриншотов LightShot. Всё началось с того, что я загрузил очередной скриншот и вспомнил недавнюю статью на хабре, где сливались...
habr.com
*** Why are there no such spiders described here?
Because working with spiders is a whole direction.
Specialization has its advantages.
*** What color should there be screenshots?
- Not so much color is important as the contrast of text and background.
*** What are the parameters for reading queries from screenshots?
- Supported formats: PNG, JPG, TIFF.
- The recommended weight of one file is not more than 2 MB.
- Not a handwritten text!
- Recommended size: from 1024 x 768 to PX 1920 x 1200 px.
*** In what countries is activity possible?
Targets for countries are acceptable:
USA, Canada, England, and also
Australia, Poland, Slovakia, Portugal, Spain, France, Denmark, Norway,
Germany, Croatia, Finland, Slovenia, Italy, Romania.
*** Data is processed only by neural network?
Tentatively data are processed:
- neural network by 70%;
- regularities by 20%;
- A person by 10%.
*** With validation of lines CC/CVV - what type does the conclusion about the quality of delivery have? What points are it contained in it?
The type of supply statistics is contained in the Example_Report_Delivery_ru.png file.
*** What is the stack?
Python, Keras, Tensorflow,
Scikit-Learn, Scipy, Opencv,
Multiprocessing, ASYNCIO, AIOHTP, AsynCPG, PSYCOPG2, PANDAS, Re,
Fastapi,
PostgreSQL, MySQL.
*** And if you try to send a malvar? ....
Save your time.
The resulting files undergo a multi -stage verification.
*** What is the time of activity?
Tentatively 9.00 - 21.00 Moscow time.
*** What is the cost of processing?
With each customer, it is specified separately.
*** Is it possible directly, without a guarantor?
All operations through the guarantor.
jabber: enterx7@xmpp.jp
element.io: https://matrix.to/#/@enterx7:matrix.org
teams.wire.: https://account.wire.com/conversati...dvu&code=UkkAVg-B9ija3kqUqaGY&domain=wire.com
qTox: 83E39F9AA994C1FF8B1316B98FF0E240D11C9B23F653A23F5A594E11B7C27915E08DCBAF6BE6
