Table 2 Efficient MOS model design... | Emerald Publishing

Table 2

Efficient MOS model design techniques.

Efficient Model Design Techniques			MOS Models
3.1	Asymmetric Convolution		DRSNet [205]
3.2	Dilated Convolution		DRSNet[205]; A Past X-shaped with CompactASPP [241]; One-Shot Animal [228]; U²- ONet [204]; Frame-Level Weakly Supervised [139]; Espnet [135]; PDB [181]; PyramidCSA [67]
3.3	Group Convolution and Depth-Wise Convolution		2D _ Separable CNN [74]; One-Shot Animal [228]; 3DS _ MM [75]; F3DsCNN [76]
3.4	Reduce Feature Map’s Volume		ChangeDet [133]; 3DPR [127]; U²-ONet [204]; AOT-T [232]; PRTM-fast [170]; FRTM [170]; G-PRTM-fast [155]; SiamMask [210]; DRSNet [205]; Adaptive Template [156]; MFCN [240]
3.5	Simplified Network Architecture		MSFgNet [160]; AGSS [112]; One-Shot Animal [228]; PiWiVOS-F [152]; FCESNet [168]; PAMINet [120]; Lightweight U-Net-like [111]; Guided Multi-Scale CNN [107]; MSCNN + Cascade [216]; Trip-Net [147]
3.6	Two-Branch Network		F3DsCNN [76]; ContextNet [166]
3.7	Channel Merging by Addition		DRSNet [205]; Lightweight U-Net-like [111]
3.8	Decoder Size Reduction		One-Shot Animal [228]; 2D _ Separable CNN [74]; 3DS _ MM [75]; F3DsCNN [76]
3.9	Quantization		T-RexNet [24]; S3-Net [38]
3.10	Small Convolution Kernel		A Past X-Shaped with CompactASPP [241]; Frame-Level Weakly Supervised [139]; MSFgNet [160]
3.11	Skip Connection		Lightweight U-Net-like [111]; BMN-BSN [142]; BSUV-Net 2.0 [193]; 3D CNN-LSTM [2]; BScGAN [10]; U²-ONet [204]; Edge Aggregation Network [157]; LSTNet [207]; Frame-Level Weakly Supervised [139]; MvRF [3]
3.12.1	Improving the Efficiency of Semi-Supervised VOS	Meta Learning	Meta-Learning [224]; e-OSVOS [137]
3.12.2		Template-Based and Propagation-Based Methods	SAT-Fast [32]; SAT [32]; Fasttmu [188]; SwiftNet [206]; RANet [219]; FRTM-fast [170];
			FRTM [170]; G-FRTM-fast [155]; TVOS [247]; TTVOS(HRNet) [154]; TTVOS(ResNet5O) [154]; GC [105]; LWL [13]; MSN [222]; RMNet [225]; SiamMask [210];
			DDEAL(Res101) [236]; LSTNet [207]; Adaptive Template [156]; AGSS [112]
3.13	Multi-Input Multi-Output Strategy		3DS _ MM [75]; F3DsCNN [76]; FCESNet [168]

Efficient Model Design Techniques			MOS Models
3.1	Asymmetric Convolution		DRSNet [205]
3.2	Dilated Convolution		DRSNet[205]; A Past X-shaped with CompactASPP [241]; One-Shot Animal [228]; U²- ONet [204]; Frame-Level Weakly Supervised [139]; Espnet [135]; PDB [181]; PyramidCSA [67]
3.3	Group Convolution and Depth-Wise Convolution		2D _ Separable CNN [74]; One-Shot Animal [228]; 3DS _ MM [75]; F3DsCNN [76]
3.4	Reduce Feature Map’s Volume		ChangeDet [133]; 3DPR [127]; U²-ONet [204]; AOT-T [232]; PRTM-fast [170]; FRTM [170]; G-PRTM-fast [155]; SiamMask [210]; DRSNet [205]; Adaptive Template [156]; MFCN [240]
3.5	Simplified Network Architecture		MSFgNet [160]; AGSS [112]; One-Shot Animal [228]; PiWiVOS-F [152]; FCESNet [168]; PAMINet [120]; Lightweight U-Net-like [111]; Guided Multi-Scale CNN [107]; MSCNN + Cascade [216]; Trip-Net [147]
3.6	Two-Branch Network		F3DsCNN [76]; ContextNet [166]
3.7	Channel Merging by Addition		DRSNet [205]; Lightweight U-Net-like [111]
3.8	Decoder Size Reduction		One-Shot Animal [228]; 2D _ Separable CNN [74]; 3DS _ MM [75]; F3DsCNN [76]
3.9	Quantization		T-RexNet [24]; S3-Net [38]
3.10	Small Convolution Kernel		A Past X-Shaped with CompactASPP [241]; Frame-Level Weakly Supervised [139]; MSFgNet [160]
3.11	Skip Connection		Lightweight U-Net-like [111]; BMN-BSN [142]; BSUV-Net 2.0 [193]; 3D CNN-LSTM [2]; BScGAN [10]; U²-ONet [204]; Edge Aggregation Network [157]; LSTNet [207]; Frame-Level Weakly Supervised [139]; MvRF [3]
3.12.1	Improving the Efficiency of Semi-Supervised VOS	Meta Learning	Meta-Learning [224]; e-OSVOS [137]
3.12.2		Template-Based and Propagation-Based Methods	SAT-Fast [32]; SAT [32]; Fasttmu [188]; SwiftNet [206]; RANet [219]; FRTM-fast [170];
			FRTM [170]; G-FRTM-fast [155]; TVOS [247]; TTVOS(HRNet) [154]; TTVOS(ResNet5O) [154]; GC [105]; LWL [13]; MSN [222]; RMNet [225]; SiamMask [210];
			DDEAL(Res101) [236]; LSTNet [207]; Adaptive Template [156]; AGSS [112]
3.13	Multi-Input Multi-Output Strategy		3DS _ MM [75]; F3DsCNN [76]; FCESNet [168]