Table 6.

Comparison of our methods based on overall classification accuracy (%) with state-of-the-art methods.

Methods	Feature size	Accuracy
Shahi et al. [31]	100-D	62.1
Basnet and Timalsina [32]	300-D	62.9
Sitaula et al. [1]	17-D	59.8
Ours (ds)	3-D	61.5
Ours (da)	17-D	59.5
Ours (ft)	300-D	68.1
Ours (ensemble)	320-D	68.7

Note. Each reported accuracy in the table is the averaged value over 10 runs on our NepCOV19Tweets dataset.