Table 3 Summary of datasets used for gesture recognition experiments.

From: A hybrid CNN-transformer framework optimized by Grey Wolf Algorithm for accurate sign language recognition

Dataset

Total number of images

Dimension

Number of classes

Class description

ASL alphabet

87,000

64 × 64

25

0–25 corresponds to A–Z

26-Space, 27-Delete, 28-Null

ASL MNIST

34,627

28 × 28

26

0–25 corresponds to A–Z

(Except class 9-J and class 25-Z due to movement gestures)

Back to article page