Text this: Computer vision-based hybrid efficient convolution for isolated dynamic sign language recognition.