The segmentation accuracy and efficiency of lightweight MOS methods on DAVIS 2016 dataset.
| No. | Year | Method | GPU | ℐ&ℱ Mean (%)↑ | ℐ Mean (%)↑ | ℱ Mean (%)↑ | Inference Speed (fps) ↑ | # Param (M) ↓ | Model Size (MB) ↓ |
|---|---|---|---|---|---|---|---|---|---|
| 1 | 2021 | AOT-T [232] | Tesla V100 | 86.8 | 86.1 | 87.4 | 51.4 | 5.7 | 23.3 |
| 2 | 2020 | FRTM-fast (ResNet-18) [170] | 78.5 | - | - | 41.3 | 13.2 | 53.9 | |
| 3 | 2020 | FRTM (Resnet-101) [170] | 83.5 | - | - | 22 | 46.9 | 187.6 | |
| 4 | 2021 | RMNet [225] | 81.5 | 80.6 | 82.3 | 12 | 53 | 202 | |
| 5 | 2020 | SAT [32] | RTX 2080 Ti | 83.1 | 82.6 | 83.6 | 39 | 72 | 288 |
| 6 | 2019 | RANet [219] | Titan Xp | 85.5 | 85.5 | 85.4 | 30.3 | 61.5 | 246 |
| 7 | 2021 | DDEAL(Res101) [236] | 85.4 | 85.1 | 85.7 | 25 | 125 | 536 | |
| 8 | 2019 | DTN [245] | GTX 1080Ti | 83.6 | 83.7 | 83.5 | 14.3 | - | - |
| 9 | 2020 | Fasttmu [188] | 78.9 | 77.5 | 80.3 | 11** | 5.2 | 37.8 | |
| 10 | 2019 | SiamMask(ResNet-50) [210] | RTX 2080 | 69.75 | 71.7 | 67.8 | 55 | 27 | 105.9 |
| 11 | 2021 | SwiftNet(ResNet-18) [206] | Tesla P100 | 90.1 | 90.3 | 89.9 | 70** | 32.5 | 130 |
| 12 | 2021 | SwiftNet(ResNet-50) [206] | 90.4 | 90.5 | 90.3 | 25** | 37.4 | 149 | |
| 13 | 2020 | GC [105] | Tesla P40 | 86.6 | 87.6 | 85.7 | 25 | 45.9 | 175 |
| 14 | 2021 | TTVOS (HRNet) [154] | - | 81.1 | - | - | 78.3 | 1.61 | 6 |
| 15 | 2021 | G-FRTM-fast (RN18, τ =1) [155] | - | 80.9 | - | - | 37.6 | - | - |
| 16 | 2020 | MSN [222] | - | 84.35 | 83.8 | 84.9 | 10 | - | - |
| No. | Year | Method | GPU | Inference Speed (fps) ↑ | # Param (M) ↓ | Model Size (MB) ↓ | |||
|---|---|---|---|---|---|---|---|---|---|
| - | - | ||||||||
| - | - | ||||||||
| - | - | ||||||||
| 11 | |||||||||
| - | - | - | |||||||
| - | - | - | - | - | |||||
| - | - | - |
Note:** The inference speeds of these models were evaluated on the DAVIS 2017 dataset in the original papers.
Sharing content requires targeting cookies to be enabled. Please update your cookie preferences to use this feature.