Semantic Segmentation on Surveillance mixture RainSnow, MTID/Drone, GRAM-RTM/M-30-HD, UT/Sherbrooke
98.5AccuracyOurs
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Oursλ1=0, λ2=0, λ3=1/2, λ4=1/2, strategy=MAP2022.11 | 98.5 | 62 | |
| Oursλ1=1/2, λ2=0, λ3=0, λ4=1/2, strategy=MAP2022.11 | 98.2 | 57.4 | |
| SM1 RainSnowλ1=1/2, λ2=0, λ3=1/2, λ4=0, strategy=MAP2022.11 | 98 | 35.3 | |
| Oursλ1=1/2, λ2=0, λ3=1/2, λ4=0, strategy=MAP2022.11 | 98 | 39.3 | |
| Oursλ1=0, λ2=1/2, λ3=0, λ4=1/2, strategy=MAP2022.11 | 98 | 60.4 | |
| Oursλ1=1/3, λ2=0, λ3=1/3, λ4=1/3, strategy=MAP2022.11 | 98 | 56.3 | |
| SM1 RainSnowλ1=1, λ2=0, λ3=0, λ4=0, strategy=MAP2022.11 | 97.9 | 35.1 | |
| Random selection of source modelsλ1=1, λ2=0, λ3=0, λ4=0, strategy=MAP2022.11 | 97.9 | 35.1 | |
| Linear combination of posteriorsλ1=1, λ2=0, λ3=0, λ4=0, strategy=MAP2022.11 | 97.9 | 35.1 | |
| Oursλ1=1, λ2=0, λ3=0, λ4=0, strategy=MAP2022.11 | 97.9 | 35.1 | |
| Oursλ1=0, λ2=1/3, λ3=1/3, λ4=1/3, strategy=MAP2022.11 | 97.8 | 57.3 | |
| Linear combination of posteriorsλ1=1/2, λ2=1/2, λ3=0, λ4=0, strategy=MAP2022.11 | 97.7 | 34.1 | |
| Oursλ1=1/3, λ2=1/3, λ3=0, λ4=1/3, strategy=MAP2022.11 | 97.7 | 55.3 | |
| Oursλ1=1/4, λ2=1/4, λ3=1/4, λ4=1/4, strategy=MAP2022.11 | 97.6 | 53.9 | |
| SM1 RainSnowλ1=1/2, λ2=1/2, λ3=0, λ4=0, strategy=MAP2022.11 | 97.4 | 35 | |
| Random selection of source modelsλ1=1/2, λ2=1/2, λ3=0, λ4=0, strategy=MAP2022.11 | 97.3 | 34 | |
| SM2 MTID/Droneλ1=1/2, λ2=1/2, λ3=0, λ4=0, strategy=MAP2022.11 | 97.2 | 33.7 | |
| Oursλ1=1/2, λ2=1/2, λ3=0, λ4=0, strategy=MAP2022.11 | 97 | 31.7 | |
| SM2 MTID/Droneλ1=1, λ2=0, λ3=0, λ4=0, strategy=MAP2022.11 | 96.4 | 26.3 | |
| SM4 UT/Sherbrookeλ1=1, λ2=0, λ3=0, λ4=0, strategy=MAP2022.11 | 96.1 | 28.3 | |
| SM3 GRAM-RTM/M-30-HDλ1=1, λ2=0, λ3=0, λ4=0, strategy=MAP2022.11 | 96 | 27.6 | |
| SM3 GRAM-RTM/M-30-HDλ1=1/2, λ2=1/2, λ3=0, λ4=0, strategy=MAP2022.11 | 95.7 | 29.6 | |
| SM4 UT/Sherbrookeλ1=1/2, λ2=1/2, λ3=0, λ4=0, strategy=MAP2022.11 | 95.5 | 28.9 |