Optical Flow Models
Model Checkpoints
The pretrained models can be downloaded by following the instructions in the official repositories RAFT and GMFlow
The links for the finetuned models are given below. Each link contains the C+B, C+T+B and C+T+S+B model checkpoints respectively.
Model Benchmarks
Pool Boiling Saturated
Model |
Method |
Chairs (Val) |
Things (Val) |
Sintel (Train) - Clean |
Sintel (Train) - Final |
KITTI (Train) - F1-EPE |
KITTI (Train) - F1-all |
Boiling (Test) |
C |
RAFT |
0.82 |
9.03 |
2.19 |
4.49 |
9.83 |
37.57 |
4.20 |
C |
GMFlow |
0.92 |
10.23 |
3.22 |
4.43 |
17.82 |
56.14 |
4.73 |
C+B |
RAFT |
0.91 |
11.22 |
2.55 |
5.16 |
13.7 |
44.44 |
2.33 |
C+B |
GMFlow |
1.31 |
11.99 |
3.78 |
5.12 |
21.91 |
63.04 |
2.36 |
C+T |
RAFT |
1.15 |
4.39 |
1.40 |
2.71 |
5.02 |
17.46 |
4.72 |
C+T |
GMFlow |
1.26 |
3.48 |
1.50 |
2.96 |
11.60 |
35.62 |
7.98 |
C+T+B |
RAFT |
1.28 |
7.69 |
1.69 |
2.95 |
9.96 |
23.61 |
2.38 |
C+T+B |
GMFlow |
1.39 |
3.88 |
1.61 |
2.91 |
14.49 |
43.09 |
2.51 |
C+T+S |
RAFT |
1.21 |
4.69 |
0.77 |
1.22 |
1.54 |
5.64 |
8.39 |
C+T+S |
GMFlow |
1.53 |
4.09 |
0.95 |
1.28 |
3.04 |
13.61 |
14.65 |
C+T+S+B |
RAFT |
1.37 |
6.59 |
0.89 |
1.60 |
1.83 |
6.44 |
2.34 |
C+T+S+B |
GMFlow |
1.65 |
4.49 |
1.07 |
1.45 |
4.06 |
18.99 |
2.56 |
Pool Boiling SubCooled
Model |
Method |
Chairs(Val) |
Things(Val) |
Sintel(Train) - Clean |
Sintel(Train) - Final |
KITTI(Train) - F1-EPE |
KITTI(Train) - F1-all |
Boiling(Test) |
C |
RAFT |
0.82 |
9.03 |
2.19 |
4.49 |
9.83 |
37.57 |
2.41 |
C |
GMFlow |
0.92 |
10.23 |
3.22 |
4.43 |
17.82 |
56.14 |
1.92 |
C+B |
RAFT |
0.89 |
11.64 |
2.58 |
5.17 |
14.01 |
47.53 |
0.63 |
C+B |
GMFlow |
1.34 |
12.23 |
3.93 |
5.27 |
22.83 |
64.41 |
0.63 |
C+T |
RAFT |
1.15 |
4.39 |
1.40 |
2.71 |
5.02 |
17.46 |
1.91 |
C+T |
GMFlow |
1.26 |
3.48 |
1.50 |
2.96 |
11.60 |
35.62 |
4.76 |
C+T+B |
RAFT |
1.24 |
6.45 |
1.54 |
2.82 |
7.35 |
20.27 |
0.70 |
C+T+B |
GMFlow |
1.46 |
4.04 |
1.66 |
3.11 |
14.50 |
44.35 |
0.67 |
C+T+S |
RAFT |
1.21 |
4.69 |
0.77 |
1.22 |
1.54 |
5.64 |
6.07 |
C+T+S |
GMFlow |
1.53 |
4.09 |
0.95 |
1.28 |
3.04 |
13.61 |
9.21 |
C+T+S+B |
RAFT |
1.35 |
6.49 |
0.87 |
1.48 |
1.79 |
6.31 |
0.64 |
C+T+S+B |
GMFlow |
1.66 |
4.43 |
1.04 |
1.42 |
3.99 |
18.83 |
0.65 |
Pool Boiling Gravity
Model |
Method |
Chairs(Val) |
Things(Val) |
Sintel(Train) - Clean |
Sintel(Train) - Final |
KITTI(Train) - F1-EPE |
KITTI(Train) - F1-all |
Boiling(Test) |
C |
RAFT |
0.82 |
9.03 |
2.19 |
4.49 |
9.83 |
37.57 |
3.30 |
C |
GMFlow |
0.92 |
10.23 |
3.22 |
4.43 |
17.82 |
56.14 |
2.40 |
C+B |
RAFT |
0.95 |
10.98 |
2.78 |
5.76 |
15.52 |
49.88 |
0.90 |
C+B |
GMFlow |
1.42 |
12.57 |
4.06 |
5.41 |
23.61 |
66.38 |
0.90 |
C+T |
RAFT |
1.15 |
4.39 |
1.40 |
2.71 |
5.02 |
17.46 |
2.75 |
C+T |
GMFlow |
1.26 |
3.48 |
1.50 |
2.96 |
11.60 |
35.62 |
3.60 |
C+T+B |
RAFT |
1.31 |
6.65 |
1.64 |
2.93 |
8.39 |
21.36 |
0.96 |
C+T+B |
GMFlow |
1.41 |
3.95 |
1.64 |
2.97 |
44.07 |
14.41 |
0.97 |
C+T+S |
RAFT |
1.21 |
4.69 |
0.77 |
1.22 |
1.54 |
5.64 |
4.74 |
C+T+S |
GMFlow |
1.53 |
4.09 |
0.95 |
1.28 |
3.04 |
13.61 |
5.49 |
C+T+S+B |
RAFT |
1.35 |
6.44 |
0.90 |
1.49 |
1.81 |
6.27 |
0.92 |
C+T+S+B |
GMFlow |
1.63 |
4.43 |
1.03 |
1.40 |
4.29 |
20.93 |
0.92 |
Flow Boiling Inlet Velocity
Model |
Method |
Chairs(Val) |
Things(Val) |
Sintel(Train) - Clean |
Sintel(Train) - Final |
KITTI(Train) - F1-EPE |
KITTI(Train) - F1-all |
Boiling(Test) |
C |
RAFT |
0.82 |
9.03 |
2.19 |
4.49 |
9.83 |
37.57 |
16.01 |
C |
GMFlow |
0.92 |
10.23 |
3.22 |
4.43 |
17.82 |
56.14 |
21.44 |
C+B |
RAFT |
1.21 |
14.99 |
3.62 |
6.78 |
22.07 |
60.64 |
10.13 |
C+B |
GMFlow |
1.50 |
13.96 |
4.52 |
5.89 |
23.45 |
65.96 |
7.01 |
C+T |
RAFT |
1.15 |
4.39 |
1.40 |
2.71 |
5.02 |
17.46 |
25.14 |
C+T |
GMFlow |
1.26 |
3.48 |
1.50 |
2.96 |
11.60 |
35.62 |
19.39 |
C+T+B |
RAFT |
1.62 |
9.09 |
2.19 |
3.77 |
13.81 |
32.13 |
9.19 |
C+T+B |
GMFlow |
1.45 |
4.15 |
1.78 |
3.05 |
15.74 |
48.34 |
7.24 |
C+T+S |
RAFT |
1.21 |
4.69 |
0.77 |
1.22 |
1.54 |
5.64 |
21.23 |
C+T+S |
GMFlow |
1.53 |
4.09 |
0.95 |
1.28 |
3.04 |
13.61 |
47.88 |
C+T+S+B |
RAFT |
2.18 |
9.32 |
1.54 |
2.90 |
2.57 |
10.93 |
9.68 |
C+T+S+B |
GMFlow |
1.72 |
4.80 |
1.18 |
1.65 |
4.69 |
24.56 |
6.88 |
Flow Boiling Gravity
Model |
Method |
Chairs(Val) |
Things(Val) |
Sintel(Train) - Clean |
Sintel(Train) - Final |
KITTI(Train) - F1-EPE |
KITTI(Train) - F1-all |
Boiling(Test) |
C |
RAFT |
0.82 |
9.03 |
2.19 |
4.49 |
9.83 |
37.57 |
20.42 |
C |
GMFlow |
0.92 |
10.23 |
3.22 |
4.43 |
17.82 |
56.14 |
14.05 |
C+B |
RAFT |
1.14 |
12.65 |
3.30 |
6.13 |
17.94 |
53.76 |
4.45 |
C+B |
GMFlow |
1.30 |
13.02 |
4.05 |
5.41 |
23.17 |
65.33 |
4.24 |
C+T |
RAFT |
1.15 |
4.39 |
1.40 |
2.71 |
5.02 |
17.46 |
21.04 |
C+T |
GMFlow |
1.26 |
3.48 |
1.50 |
2.96 |
11.60 |
35.62 |
13.37 |
C+T+B |
RAFT |
1.41 |
6.85 |
1.97 |
3.89 |
10.96 |
27.82 |
4.24 |
C+T+B |
GMFlow |
1.37 |
3.89 |
1.71 |
3.05 |
14.29 |
45.03 |
3.94 |
C+T+S |
RAFT |
1.21 |
4.69 |
0.77 |
1.22 |
1.54 |
5.64 |
21.04 |
C+T+S |
GMFlow |
1.53 |
4.09 |
0.95 |
1.28 |
3.04 |
13.61 |
22.84 |
C+T+S+B |
RAFT |
1.59 |
7.29 |
1.13 |
2.14 |
2.22 |
9.34 |
4.12 |
C+T+S+B |
GMFlow |
1.62 |
4.42 |
1.04 |
1.43 |
3.86 |
17.76 |
3.93 |
SciML Models
Model Checkpoints
The model checkpoints are organized based on the dataset they were trained on. All model weights are included in
the .tar.gz
.
Model Benchmarks
Pool Boiling Subcooled
|
UNet-bench |
UNet-mod |
FNO |
UNO |
FFNO |
GFNO |
Rel. Err. |
0.036 |
0.051 |
0.052 |
0.062 |
0.050 |
0.071 |
RMSE |
0.035 |
0.049 |
0.050 |
0.059 |
0.048 |
0.068 |
BRMSE |
0.073 |
0.146 |
0.118 |
0.145 |
0.124 |
0.209 |
IRMSE |
0.113 |
0.157 |
0.189 |
0.220 |
0.155 |
0.290 |
Max Err. |
2.1 |
1.937 |
2.32 |
2.659 |
2.21 |
2.882 |
F. Low |
0.281 |
0.348 |
0.373 |
0.440 |
0.472 |
0.557 |
F. Mid |
0.266 |
0.386 |
0.390 |
0.453 |
0.375 |
0.547 |
F. High |
0.040 |
0.059 |
0.055 |
0.068 |
0.057 |
0.090 |
Pool Boiling Saturated
|
UNet-bench |
UNet-mod |
FNO |
UNO |
FFNO |
GFNO |
Rel. Err. |
0.035 |
0.039 |
0.052 |
0.072 |
0.053 |
0.066 |
RMSE |
0.035 |
0.039 |
0.052 |
0.071 |
0.052 |
0.065 |
BRMSE |
0.082 |
0.147 |
0.165 |
0.214 |
0.163 |
0.218 |
IRMSE |
0.052 |
0.083 |
0.125 |
0.116 |
0.095 |
0.152 |
Max Err. |
1.701 |
1.785 |
2.055 |
2.23 |
1.788 |
1.793 |
F. Low |
0.257 |
0.241 |
0.373 |
0.568 |
0.416 |
0.463 |
F. Mid |
0.268 |
0.273 |
0.377 |
0.532 |
0.399 |
0.496 |
F. High |
0.023 |
0.042 |
0.053 |
0.069 |
0.052 |
0.070 |
Pool Boiling Gravity
|
UNet-bench |
UNet-mod |
FNO |
UNO |
FFNO |
GFNO |
Rel. Err. |
0.042 |
0.051 |
0.062 |
0.081 |
0.061 |
0.124 |
RMSE |
0.040 |
0.049 |
0.062 |
0.077 |
0.058 |
0.118 |
BRMSE |
0.070 |
0.139 |
0.150 |
0.184 |
0.131 |
0.642 |
IRMSE |
0.108 |
0.201 |
0.266 |
0.335 |
0.239 |
0.424 |
Max Err. |
2.92 |
2.846 |
3.626 |
3.667 |
3.22 |
4.0 |
F. Low |
0.491 |
0.466 |
0.500 |
0.746 |
0.578 |
1.095 |
F. Mid |
0.275 |
0.326 |
0.440 |
0.548 |
0.423 |
0.760 |
F. High |
0.033 |
0.049 |
0.054 |
0.077 |
0.049 |
0.171 |
Flow Boiling Gravity
|
UNet-bench |
UNet-mod |
FNO |
UNO |
Rel. Err. |
0.055 |
0.095 |
0.134 |
0.157 |
RMSE |
0.051 |
0.088 |
0.123 |
0.144 |
BRMSE |
0.080 |
0.240 |
0.267 |
0.301 |
IRMSE |
0.091 |
0.214 |
0.351 |
0.379 |
Max Err. |
2.560 |
2.800 |
3.951 |
3.990 |
F. Low |
0.263 |
0.388 |
0.591 |
0.602 |
F. Mid |
0.281 |
0.419 |
0.551 |
0.692 |
F. High |
0.149 |
0.276 |
0.389 |
0.449 |
Flow Boiling Inlet Velocity
|
UNet-bench |
UNet-mod |
FNO |
UNO |
Rel. Err. |
0.106 |
0.122 |
0.202 |
0.225 |
RMSE |
0.097 |
0.111 |
0.184 |
0.205 |
BRMSE |
0.271 |
0.315 |
0.486 |
0.471 |
IRMSE |
0.178 |
0.193 |
0.252 |
0.319 |
Max Err. |
2.900 |
2.862 |
3.527 |
3.992 |
F. Low |
0.571 |
0.584 |
1.061 |
0.981 |
F. Mid |
0.511 |
0.566 |
1.084 |
1.334 |
F. High |
0.264 |
0.315 |
0.494 |
0.523 |
Velocity + Temp: Pool Boiling Subcooled
|
Metric |
UNet-bench |
UNet-mod |
P-UNet-mod |
FNO |
UNO |
FFNO |
Temp. |
Rel. Err. |
0.106 |
0.074 |
0.040 |
0.066 |
0.322 |
0.067 |
|
RMSE |
0.097 |
0.072 |
0.039 |
0.064 |
0.051 |
0.065 |
|
BRMSE |
0.271 |
0.211 |
0.113 |
0.142 |
0.050 |
0.237 |
|
IRMSE |
0.178 |
0.191 |
0.129 |
0.212 |
0.077 |
0.201 |
|
Max Err. |
2.83 |
2.340 |
2.741 |
2.566 |
7.492 |
2.488 |
|
F. Low |
1.407 |
0.495 |
0.244 |
0.531 |
4.902 |
0.498 |
|
F. Mid |
0.740 |
0.583 |
0.280 |
0.518 |
1.594 |
0.507 |
|
F. High |
0.090 |
0.076 |
0.051 |
0.061 |
0.442 |
0.083 |
$x$ Vel. |
Rel. Err. |
0.681 |
0.553 |
0.429 |
0.642 |
1.834 |
0.547 |
|
RMSE |
0.021 |
0.015 |
0.012 |
0.018 |
0.051 |
0.015 |
|
BRMSE |
0.023 |
0.023 |
0.022 |
0.023 |
0.050 |
0.024 |
|
IRMSE |
0.051 |
0.049 |
0.043 |
0.049 |
0.077 |
0.050 |
|
Max Err. |
0.294 |
0.320 |
0.353 |
0.32 |
0.574 |
0.334 |
|
F. Low |
0.329 |
0.225 |
0.142 |
0.258 |
0.677 |
0.189 |
|
F. Mid |
0.115 |
0.105 |
0.079 |
0.110 |
0.328 |
0.107 |
|
F. High |
0.012 |
0.012 |
0.011 |
0.012 |
0.092 |
0.012 |
$y$ Vel. |
Rel. Err. |
0.563 |
0.400 |
0.311 |
0.415 |
1.170 |
0.397 |
|
RMSE |
0.021 |
0.015 |
0.012 |
0.016 |
0.044 |
0.015 |
|
BRMSE |
0.015 |
0.012 |
0.010 |
0.011 |
0.035 |
0.013 |
|
IRMSE |
0.046 |
0.046 |
0.038 |
0.045 |
0.066 |
0.047 |
|
Max Err. |
0.292 |
0.272 |
0.280 |
0.290 |
0.288 |
0.289 |
|
F. Low |
0.387 |
0.190 |
0.132 |
0.180 |
0.660 |
0.176 |
|
F. Mid |
0.128 |
0.112 |
0.085 |
0.111 |
0.240 |
0.106 |
|
F. High |
0.011 |
0.010 |
0.009 |
0.010 |
0.067 |
0.010 |