Automated analysis of pectoralis major thickness in pec-fly exercises: evolving from manual measurement to deep learning techniques

Visual Computing for Industry, Biomedicine, and Art

Table 1 Performance comparison of RNN and two-layer CNN models

	Smooth loss	MSE loss	Model parameter size
RNN (LSTM) [22]	0.1004 ± 0.0004	0.0016 ± 0.0004	21,223,410
Two-layer CNN (our)	0.1005 ± 0.0001	0.0015 ± 0.0002	21,210,289