FlowSeek
FlowSeek (University of Bologna, 2025), size T: SEA-RAFT optical flow that also uses a frozen Depth Anything V2 Small: its features help the matching and its depth tells which flow a camera motion would cause. Apache-2.0 code; the authors' weights carry no separate licence.
experimental Fork of mattpoggi/flowseek@ba64073 · backends: portable, kernels
- Paper FlowSeek: Optical Flow Made Easier with Depth Foundation Models and Motion Bases ICCV 2025 · first released 5 Sept 2025
Paper 2025 Added in 0.2.0 Evaluated with 0.2.0
Use it
pip install fastforks Lean install with only what this model needs: pip install fastforks-core
from fastforks import flow
model = flow.load("flowseek-t") # kernels on a supported GPU, else portable PyTorch
result = model(frame1, frame2)
result.flow
result.backend # what ran: mode, device, dtype, kernels fastforks bench flowseek-t Times the model on this machine. load(…, mode="portable") or mode="kernels" picks a backend; usage guide.
Speed and memory
Every backend on one GPU (RTX 5060 Ti 16 GB) and the same input, the authors’ code included. Kernels runs fp16. Portable runs fp16 on CUDA.
| Backend | dtype | Time per call (ms) | vs original fp16 | Peak memory allocated / reserved | Load (s) | First call (s) |
|---|---|---|---|---|---|---|
| Original fp32upstream FlowSeek (flowseek-T.json), iters=4 | fp32 | 650 | 0.70× | 9,902 MiB/10,052 MiB | 2.6 | 1.03 |
| Original fp16torch.autocast(float16) around the call; upstream FlowSeek (flowseek-T.json), iters=4 | fp16 | 457 | – | 8,693 MiB/9,880 MiB | 0.9 | 0.63 |
| Portable fp32 | fp32 | not measured | – | – | – | – |
| Portable | fp16 | 395 | 1.2× | 8,671 MiB/8,712 MiB | 0.2 | 0.40 |
| Kernelson-demand correlation; code snapshot was measured before a style-only constant rename; 1080p improves speed and memory, KITTI-sized runtime is similar to portable | fp16 | 327 | 1.4× | 2,027 MiB/2,192 MiB | 0.3 | 0.33 |
| Portable + torch.compile | 258 | 1.8× | 6,126 MiB/6,212 MiB | 54 (build) | – | |
| TensorRT fp16 | fp16 | 168 | 2.7× | 10.8 GiBenginenot comparable | 107 (build) | – |
Measured on RTX 5060 Ti 16 GB, torch 2.14.1+cu130, 2026-11-11 to 2026-11-17. First call, 3 warm-up, peak reset, median of 3 rounds x median of 5. Ratios divide two
times of this table (same GPU, same input). Original fp16 is the authors’ own precision option, else torch.autocast. torch.compile and TensorRT: build time instead of load time. TensorRT memory (engine, not comparable) is the GPU memory the engine took, read from the driver because TensorRT allocates outside PyTorch, so it can't be compared with the PyTorch peaks of the other rows.
Other devices
| Device | Backend | dtype | ms / call | Peak memory | Output vs reference | Input |
|---|---|---|---|---|---|---|
| Apple M1 Pro 16 GB MPS, torch 2.14.1, 2026-11-14 | Original | fp32 | out of memory | – | 1080x1920 | |
| Portable | fp16 | out of memory | – | 1080x1920 | ||
| Portable | fp32 | out of memory | – | 1080x1920 | ||
| Kernels | fp32 | not implemented | – | 1080x1920 | ||
CUDA memory is torch’s peak allocated / reserved; Apple has no such counter, so the Mac rows show the process footprint (weights, Python and torch included). “Output vs reference” compares the device’s output with the stored golden output under the card’s tolerance; – = not checked on that device. Kernels “not implemented” = no kernels for this model on that backend yet: it runs portable there.
Licence and weights
Commercial use allowed
- Code
- Apache-2.0 AND BSD-3-Clause LICENSELICENSE-BSD-3-Clause
- Weights
- Apache-2.0 AND BSD-3-Clause (assumed from code)(not stated by the authors; assumed to follow the code) source
- Training data
-
- TartanAir CC-BY-4.0
- FlyingChairs (Freiburg) research-only, non-commercial
- Freiburg SceneFlow (FlyingThings3D, Driving, Monkaa) research-only, non-commercial
- MPI Sintel optical flow noncommercial, non-commercial
- KITTI stereo / optical flow 2012 / 2015 CC-BY-NC-SA-3.0, non-commercial
- HCI HD1K optical flow benchmark not stated
- weights licence not stated by the original authors; assumed to be the code licence
Our reading of the licences, not legal advice.
Weights
flowseek_T_TartanCT_TSKH.pth 171 MB· hosted on Google Drivefastforks downloads and checks these files on first use (fastforks download flowseek-t).
Accuracy
The test set through the authors’ code and through ours: same checkpoint, same protocol. The distance between the original’s fp32 and fp16 runs is what half precision alone costs; our runs should stay within it.
KITTI 2015 train
EPE / Fl-all %trained on this dataset’s public images: not zero-shot
| Run | dtype | EPE ↓ | Fl-all % ↓ | Δ EPE vs fp32 original | pairs | ms / pair |
|---|---|---|---|---|---|---|
| Original fp32 | fp32 | 1.260 | 3.88 | reference | 200 | 149 |
| Original fp16 | fp16 | 1.360 | 4.35 | −0.099 | 200 | 84.8 |
| Portable fp32 | fp32 | 1.259 | 3.88 | +0.001 | 200 | 154 |
| Portable | fp16 | 1.348 | 4.33 | −0.088 | 200 | 81.0 |
| Kernels | not measured | |||||
Frames _10 -> _11 at full size against the flow_occ ground truth; EPE = mean of the per-pair mean end-point error, Fl-all = share of all valid pixels with an error > 3 px and > 5 % of the GT magnitude. Original = the authors’ code, same checkpoint and protocol. Measured on original fp32, original fp16: RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-11; portable fp32: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 6f0a70e1, 2026-11-17; portable: RTX 5060 Ti, torch 2.14.1+cu130, fastforks c4a67979, 2026-11-11.
Closeness to the original
Beyond the test-set scores: the raw output of ours against the original’s fp32 output on the same input, pixel by pixel. CPU, float32: max |diff| <= 0.001 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.0077 and p99 <= 0.041 (absolute error of the output flow).
| Check | Tolerance | Measured |
|---|---|---|
| CPU fp32, max |diff| | ≤ 0.001 | not recorded |
| GPU default dtypes, mean |diff| | ≤ 0.0077 | 0.00508 |
| GPU default dtypes, p99 |diff| | ≤ 0.041 | 0.0272 |
Recorded for flowseek-t on RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-17, with these kernels: raft_flow
Details
FlowSeek (Poggi & Tosi 2025), size T: SEA-RAFT (ResNet18 encoders, initial flow regression, ConvNeXt update block, 4 iterations) helped by a frozen Depth Anything V2 Small. Its features join the matching features, and its relative depth gives 8 motion bases (the flow fields that camera translations and rotations would cause), which a third encoder turns into extra context. Each image is resized to 518 x 518 for Depth Anything, whatever the input size. The authors' released weights; the network is ported from mattpoggi/flowseek. `iters=12` is the paper's FlowSeek (S). The larger M and L sizes use Depth Anything V2 Base, whose weights are non-commercial, and are not included.
- Input / output
- Input: two RGB images (consecutive frames). Output: per-pixel motion from the first to the second (dx, dy in pixels).
- Input size
- any H x W >= 128, resized or padded inside the model; tested up to 1024x2048 (8,850 MiB allocated)
- Options
iters=4>= 1· changes the output
- Outputs
flow- CUDA kernels
raft_flowRAFT optical flow kernels
- Apple (MPS)
- fp32, fp16
Full reference, generated from the repository
FlowSeek (University of Bologna, 2025), size T: SEA-RAFT optical flow that also uses a frozen Depth Anything V2 Small: its features help the matching and its depth tells which flow a camera motion would cause. Apache-2.0 code; the authors' weights carry no separate licence.
Commercial use: yes.
| variant | input | backend | time (ms) | peak memory (MiB, allocated / reserved) |
|---|---|---|---|---|
flowseek-t | synthetic (1080x1920) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 650 457 395 327 258 168 | 9902 / 10052 8693 / 9880 8671 / 8712 2027 / 2192 6126 / 6212 11328 / 11350 |
flowseek-t-things | synthetic (1080x1920) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 650 458 395 327 258 168 | 9902 / 10012 8693 / 9820 8671 / 8712 2027 / 2192 6126 / 6212 11330 / 11352 |
flowseek-t-things-notartan | synthetic (1080x1920) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 651 458 396 328 258 168 | 9902 / 10052 8693 / 9820 8671 / 8712 2027 / 2192 6126 / 6212 11328 / 11350 |
Details
- Original code: https://github.com/mattpoggi/flowseek at
ba6407370bce - Code licence: Apache-2.0 AND BSD-3-Clause (permissive); files:
LICENSE,LICENSE-BSD-3-Clause - Weights licence: Apache-2.0 AND BSD-3-Clause (assumed from code)
- Licence evidence: https://github.com/mattpoggi/flowseek/blob/ba6407370bceeb96244cad4d60d37489f6eb471e/scripts/get_weights.sh
- Training data: TartanAir, FlyingChairs (Freiburg), Freiburg SceneFlow (FlyingThings3D, Driving, Monkaa), MPI Sintel optical flow, KITTI stereo / optical flow 2012 / 2015, HCI HD1K optical flow benchmark (TartanAir pretraining, FlyingChairs, FlyingThings3D (part of SceneFlow), then a mix of FlyingThings3D, Sintel, KITTI 2015 and HD1K (TSKH); the Depth Anything V2 Small inside was trained by its authors on synthetic and pseudo-labelled real images)
flowseek-t-things: TartanAir, FlyingChairs (Freiburg), Freiburg SceneFlow (FlyingThings3D, Driving, Monkaa) (TartanAir, FlyingChairs, then FlyingThings3D (part of SceneFlow))flowseek-t-things-notartan: FlyingChairs (Freiburg), Freiburg SceneFlow (FlyingThings3D, Driving, Monkaa) (FlyingChairs, then FlyingThings3D (part of SceneFlow))
- Options:
iters= 4 (>= 1) - Kernel libraries:
raft_flow - Variants:
flowseek-t,flowseek-t-things,flowseek-t-things-notartan - Outputs: flow
- Input size: any H x W >= 128, resized or padded inside the model
- Tested up to (portable, fp16, RTX 5060 Ti 16 GB):
flowseek-t1024x2048 (8.6 GiB peak allocated, 10.2 reserved);flowseek-t-things1024x2048 (8.6 GiB peak allocated, 10.2 reserved);flowseek-t-things-notartan1024x2048 (8.6 GiB peak allocated, 10.2 reserved) flowseek-t1080x1920, first call: portable + torch.compile 54 s (cold Inductor cache), TensorRT fp16 107 s (ONNX + engine).flowseek-t-things1080x1920, first call: portable + torch.compile 55 s (cold Inductor cache), TensorRT fp16 102 s (ONNX + engine).flowseek-t-things-notartan1080x1920, first call: portable + torch.compile 54 s (cold Inductor cache), TensorRT fp16 98 s (ONNX + engine).flowseek-toriginal (fp32) 1080x1920: upstream FlowSeek (flowseek-T.json), iters=4flowseek-toriginal 1080x1920: torch.autocast(float16) around the call; upstream FlowSeek (flowseek-T.json), iters=4flowseek-tkernels 1080x1920: on-demand correlation; code snapshot was measured before a style-only constant rename; 1080p improves speed and memory, KITTI-sized runtime is similar to portableflowseek-t-thingsoriginal (fp32) 1080x1920: upstream FlowSeek (flowseek-T.json), iters=4flowseek-t-thingsoriginal 1080x1920: torch.autocast(float16) around the call; upstream FlowSeek (flowseek-T.json), iters=4flowseek-t-thingskernels 1080x1920: on-demand correlation; code snapshot was measured before a style-only constant rename; 1080p improves speed and memory, KITTI-sized runtime is similar to portableflowseek-t-things-notartanoriginal (fp32) 1080x1920: upstream FlowSeek (flowseek-T.json), iters=4flowseek-t-things-notartanoriginal 1080x1920: torch.autocast(float16) around the call; upstream FlowSeek (flowseek-T.json), iters=4flowseek-t-things-notartankernels 1080x1920: on-demand correlation; code snapshot was measured before a style-only constant rename; 1080p improves speed and memory, KITTI-sized runtime is similar to portableflowseek-tload / first call: original 0.87 s / 0.629 s, portable 0.18 s / 0.458 s (375x1242), kernels 0.28 s / 0.382 s (375x1242)flowseek-t-thingsload / first call: original 0.78 s / 0.459 s, portable 0.18 s / 0.469 s (375x1242), kernels 0.67 s / 0.404 s (375x1242)flowseek-t-things-notartanload / first call: original 0.81 s / 0.459 s, portable 0.18 s / 0.46 s (375x1242), kernels 0.37 s / 0.383 s (375x1242)- Protocol, first call, 3 warm-up, peak reset, median of 3 rounds x median of 5
Measured
Frames _10 -> _11 at full size against the flow_occ ground truth; EPE = mean of the per-pair mean end-point error, Fl-all = share of all valid pixels with an error > 3 px and > 5 % of the GT magnitude. Original = the original code named under Details, same checkpoint and protocol.
| variant | dataset | original fp32 | original fp16 | portable | portable fp32 | note |
|---|---|---|---|---|---|---|
flowseek-t | KITTI 2015 train, EPE / Fl-all % | 1.260 / 3.88 | 1.360 / 4.35 | 1.348 / 4.33 | 1.259 / 3.88 | trained on this dataset's public images; not a zero-shot result |
flowseek-t-things | KITTI 2015 train, EPE / Fl-all % | 4.121 / 12.35 | 4.153 / 12.25 | 4.143 / 12.27 | 4.098 / 12.34 | |
flowseek-t-things-notartan | KITTI 2015 train, EPE / Fl-all % | 3.923 / 12.83 | 3.904 / 12.97 | 3.933 / 12.97 | 3.925 / 12.81 |
Measured on portable: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks c4a67979, 2026-11-11; portable32: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 6f0a70e1, 2026-11-17.
Parity
CPU, float32: max |diff| <= 0.001 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.0077 and p99 <= 0.041 (absolute error of the output flow).
| variant | kernels error vs original, mean / p99 | GPU, torch | kernels used |
|---|---|---|---|
flowseek-t | 0.00508 / 0.0272 | RTX 5060 Ti, torch 2.14.1+cu130 | raft_flow@a101a33607b6:sm_120 |
flowseek-t-things | 0.00427 / 0.0176 | RTX 5060 Ti, torch 2.14.1+cu130 | raft_flow@a101a33607b6:sm_120 |
flowseek-t-things-notartan | 0.00227 / 0.01045 | RTX 5060 Ti, torch 2.14.1+cu130 | raft_flow@a101a33607b6:sm_120 |
Example output
One fixed input through this checkpoint, the same for every model of the task (portable · float16 · NVIDIA GeForce RTX 5060 Ti).
Input: Motorcycle, Middlebury 2014 stereo (left and right view, downscaled 4x), Scharstein et al., Middlebury College (research dataset, no licence stated).