DEVELOPMENT OF A MODEL FOR REAL-TIME RECOGNITION OF KAZAKH SIGN LANGUAGE USING MEDIAPIPE AND DEEP LEARNING METHODS
Bibliographic record
Abstract
Technologies for automatic processing of sign language have become an urgent need for members of society with hearing and speech impairments who face inequality in the era of digital transformation. In recent years, the issue of considering sign language as a formal structure equal to natural language and adapting it to automatic systems has attracted increasing attention from researchers. To perform the task of automatically translating information from natural language into sign language, glosses, which are the textual representation of sign language, are used as an intermediate layer. For this purpose, this study proposes a new method for converting Kazakh language text, which reflects the morphological features of the Kazakh language, into sign language glosses using natural language processing techniques. In particular, a Seq2Seq architecture based on the ByT5 small model is applied. The obtained results demonstrate that the generated gloss sequences are compact and semantically rich while preserving the internal structure of sign language. The gloss sequence makes it possible to automate the work of an interpretable intermediate layer that represents sign language movements as logical units similar to written language. The transformed gloss sequence preserves the structure of sign language, reduces redundancy, and improves sentence coherence. Thus, the use of only semantically meaningful units to control sign language avatars reduces computational requirements. Short and semantically rich glosses serve as an effective resource for synthesizing hand movements in sign language.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".