Downloads · 30 days
16
7% of all-time downloads
mlx-community/SEA-RAFT-Tartan-C-T-TSKH-spring540x960-S-mlx
SEA-RAFT-Tartan-C-T-TSKH-spring540x960-S-mlx is a image-to-image model from mlx-community. Use it when you need one image transformed into another. It is set up for mlx. The card lists the license as bsd-3-clause.
SEA-RAFT-S (Wang, Lipson, Deng — "SEA-RAFT: Simple, Efficient, Accurate RAFT for Optical Flow", ECCV 2024) converted to Apple MLX format (NHWC), consumed by xocialize/sea-raft-mlx (Python) and the upcoming MLX-Swift p…
Downloads · 30 days
16
7% of all-time downloads
All-time downloads
227
Public
Parameters
8.9M
35.6 MB on disk
Likes
3
Public
Click a slice to open those files.
.safetensors35.6 MB · 100%
From the Hugging Face model README
SEA-RAFT-S (Wang, Lipson, Deng — "SEA-RAFT: Simple, Efficient, Accurate RAFT for Optical
Flow", ECCV 2024) converted to Apple MLX format (NHWC), consumed by
xocialize/sea-raft-mlx (Python) and the upcoming
MLX-Swift port.
This is the full training-schedule checkpoint (best accuracy): TartanAir → FlyingChairs →
FlyingThings3D → Sintel/KITTI/HD1K → Spring. Note: the later stages use research-restricted
datasets (documented here per provenance policy). For the cleanest chain (TartanAir/CC-BY only), see SEA-RAFT-Tartan480x640-S-mlx.
Converted weights verified against the PyTorch reference (princeton-vl/SEA-RAFT@9137517) on an
identical input pair: final flow cosine 1.0000, max_abs 1.2e-3 px, EPE 1.1e-4 px (fp32 CPU).
[N, H, W, 3] RGB in 0–255; output [N, H, W, 2] flow in pixels.model.safetensors (MLX NHWC, fp32) · config.json.
MemorySlices/Tartan-C-T-TSKH-spring540x960-S
(tagged bsd-3-clause).