Table 4.

Comparison of the performance of the sensor-based and vision-based approaches.

Ref.	Asana (No.)	Method	Performance	Method
[86]	6	CNN, LSTM	99.91%	Sensor-based
[89]	6	CNN, LSTM	89.29%, 96.31%
[90]	5	CNN	96%
[88]	10	XBoost	99.2%
[124]	10	RF, KNN, SVM, DT	94.28%	Vision-based
[14]	26	DCNN	99.99%
[110]	42	CNN	98.93%
[99]	5	Adaboost	94.78%
[101]	10	RF	96.47%
[103]	4	SVM	94.9%
[3]	4	KNN	93.1%
[114]	17	LSTM	97.7%
[105]	8	CNN, SAE	90%, 70%