Visual Place Recognition on General Efficiency Benchmark
2.4Params (M)Text2Graph VPR
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Text2Graph VPRBackbone=GAT, Descriptor Dim.=10242025.12 | 2.4 | 2.1 | 0.4 | 1 | |
| SeqVLADBackbone=ResNet-18, Descriptor Dim.=40962025.12 | 7.15 | 17.2 | 1.1 | 2.19 | |
| NetVLADBackbone=ResNet-18, Descriptor Dim.=40962025.12 | 11.7 | 30.4 | 1.1 | 2.26 | |
| JistBackbone=ResNet-18, Descriptor Dim.=5122025.12 | 11.7 | 11.1 | 0.1 | 2.04 | |
| Spatio-Temporal-SeqVPRBackbone=CCT384, Descriptor Dim.=40962025.12 | 13.06 | 18.7 | — | — | |
| SeqMatchBackbone=VGG-16, Descriptor Dim.=40962025.12 | 14.74 | 39.8 | 500.1 | 2.68 | |
| Delta descriptorsBackbone=VGG-16, Descriptor Dim.=40962025.12 | 14.74 | 39.8 | 1.1 | 2.68 | |
| HVPRBackbone=VGG-16, Descriptor Dim.=40962025.12 | 14.74 | 39.8 | 0.9 | 2.68 | |
| SeqNetBackbone=VGG-16, Descriptor Dim.=40962025.12 | 83.89 | 31 | 1.1 | 2.68 |