Record source

Russian model registry

The named source does not confirm ownership or quality.

Open source data

Technical Score

Not enough data to calculate

Coverage 0% · technical-v2

How the score is calculated

User Rating

There are no published reviews yet

Published reviews are not mixed with technical observations.

Go to reviews

Latest endpoint observation

There is no endpoint observation yet

The status applies only to the checked endpoint, not the whole service.

Open history

About this service

What it does

Technical interface from registry.cit.gov.ru. Check the source for its exact purpose.

Task category

Models & APIs

Named provider

registry.cit.gov.ru

How to connect

Download

User reviews

There are no published reviews yet

New reviews are not being accepted yet.

Technical observation history

Recent endpoint checks

No check history has been published yet.

Problems in the observation history

There are no problems in the available history.

Russian market

Only values and evidence present in the current record are shown.

Market in the source record: Russia
Ruble payments: Unknown
Personal data in Russia: Unknown

Endpoints and technical capabilities

Each observation applies to a specific endpoint.

The record has no endpoint available for technical checking.

Access methods: Download
Protocols: Not stated in the current record
Version: 1
Technical record name: XCT-SAM

Source data

This data comes from the named source. It is not proof of ownership or an independent quality assessment.

Original description

Исследовательская разработка (Research Preview), представляющая собой фреймворк для параметро-эффективной доменной адаптации модели Segment Anything Model (SAM) под задачи сегментации внутренних дефектов (пористость, трещины) в данных рентгеновской компьютерной томографии (XCT). Архитектура использует метод сверточной низкоранговой адаптации (Conv-LoRA) внедренный в энкодер ViT-H, что позволяет сохранять пространственные признаки при обучении на малых выборках. Модель применяет стратегию последовательной адаптации (Sequential Domain Adaptation): первичная настройка на синтетических данных (CycleGAN-XCT) с последующей донастройкой на реальных промышленных сканах (NIST).

Open the named source record

Record source: Russian model registry
Source updated: 28 Jul 2026, 11:27 UTC
Ownership status: Owner not confirmed
Source identifier: ru_ml_registry

Corrections and future affiliation verification

A public record can be corrected through support without hiding its source or observation history. Affiliation verification and a formal dispute flow have not launched yet.

Report an error How service affiliation will be verified Contact support about a disputed record

Snapshot received: 9 Aug 2026, 20:01 UTC.