LoMa
LoMa (Nordström, Edstedt et al. 2026): sparse matching of two photos of one scene: DaD keypoints, DeDoDe-G descriptions (VGG19 plus DINOv2) and a light LightGlue-style transformer matcher; matches as keypoint index pairs (the form COLMAP takes). MIT code and weights; DINOv2 Apache-2.0.
experimental Fork of davnords/loma@ecc947d · backends: portable, kernels
- Paper LoMa: Local Feature Matching Revisited ECCV 2026 · first released 6 Apr 2026
Paper 2026 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 matching
model = matching.load("loma-b") # kernels on a supported GPU, else portable PyTorch
result = model(image_a, image_b)
result.keypoints1 # also: keypoints2, pairs, scores
result.backend # what ran: mode, device, dtype, kernels
points1, points2 = result.matches(5000) # matched pixels, sampled where the model is certain
pose = matching.estimate_pose(points1, points2, K1, K2) # relative pose on the GPU (RANSAC); K = intrinsics [3, 3]
pose.R, pose.t, pose.inliers # rotation, unit translation, inlier mask fastforks bench loma-b 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 bf16 on CUDA.
Input: synthetic, 768x1024 (the task’s standard size)
| Backend | dtype | Time per call (ms) | vs original fp16 | Peak memory allocated / reserved | Load (s) | First call (s) |
|---|---|---|---|---|---|---|
| Original fp32 | fp32 | not measured | – | – | – | – |
| Original (own default)upstream LoMa(LoMaB(mp=False)), float32, 2048 keypoints | fp32 | 1036 | – | 4,176 MiB/4,390 MiB | 8.5 | 1.37 |
| Portable fp32 | fp32 | not measured | – | – | – | – |
| Portable | bf16 | 308 | 3.4× | 1,767 MiB/1,878 MiB | 2.5 | 1.27 |
| Kernels | not measured | – | – | – | – | |
| Portable + torch.compile | 285 | 3.6× | 2,870 MiB/3,216 MiB | 104 (build) | – | |
| TensorRTplanned for 0.2.1 | fails | – | – | – | – |
Measured on RTX 5060 Ti 16 GB, torch 2.14.1+cu130, 2026-11-15. 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, 2026-11-17 | Portable | fp32 | 3037 | 8,147 MiB process | skip | 768x1024 |
| Kernels | fp32 | not implemented | – | 768x1024 | ||
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
- MIT AND Apache-2.0 LICENSELICENSE-DADLICENSE-DEDODELICENSE-DINOv2
- Weights
- MIT AND Apache-2.0 (assumed from code)(not stated by the authors; assumed to follow the code) source
- Training data
- weights licence not stated by the original authors; assumed to be the code licence
Our reading of the licences, not legal advice.
Weights
loma_B.pt 758 MB· hosted on GitHubfastforks downloads and checks these files on first use (fastforks download loma-b).
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.
MegaDepth-1500
every 5th pair, relative pose AUC@5 / @10 / @20 degrees
| Run | dtype | AUC@5° ↑ | AUC@10° ↑ | AUC@20° ↑ | Δ AUC@5° vs fp32 original | pairs | ms / pair |
|---|---|---|---|---|---|---|---|
| Original fp32 | fp32 | 54.5 | 71.6 | 83.7 | reference | 300 | 1033 |
| Original fp16 | not measured | ||||||
| Portable fp32 | not measured | ||||||
| Portable | bf16 | 56.2 | 72.6 | 83.9 | +1.7 | 300 | 403 |
| Kernels | not measured | ||||||
Dense matchers: 5000 matches per pair sampled from the warp by certainty (upstream's balanced KDE sampling); sparse matchers: all their matches of one call (LightGlue, XFeat: LightGlue's MegaDepth-1500 setting; LoFTR, DeDoDe: their card defaults). OpenCV essential-matrix RANSAC (0.5 px threshold, normalised by the mean focal length); AUC of the larger of the rotation and translation-direction errors. The same 300 pairs for all modes, so float32 and bfloat16 compare pair for pair. Original = the authors’ code, same checkpoint and protocol. Measured on original fp32: RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-16; portable: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 68b2f345, 2026-11-16.
Relative pose on the GPU
A dense matcher is usually followed by RANSAC on its matches to get the camera motion. fastforks does that step on the
GPU too: result.matches(n) samples matched points where the model is certain, and matching.estimate_pose runs essential-matrix RANSAC on them, many image pairs per call, with no copy to
the CPU. Like the models it has a portable (plain PyTorch) and a kernels (CUDA) backend.
| Solver | Runs on | Threshold (px) | Pose AUC@5 / @10 / @20 | Pairs per second |
|---|---|---|---|---|
| fastforks estimate_pose, kernels | GPU, batches of 500 pairs | 1.0 | 69.9 / 81.5 / 89.3 | 18,446 |
| fastforks estimate_pose, portable | GPU, batches of 500 pairs | 1.0 | 69.6 / 81.2 / 89.1 | 418 |
| PoseLib 2.0.5 | CPU, 16 processes | 1.0 | 70.0 / 81.6 / 89.5 | 325 |
| OpenCV USAC (MAGSAC++) | CPU, 16 processes | 0.5 | 65.7 / 78.6 / 87.3 | 271 |
| OpenCV RANSAC | CPU, 16 processes | 0.5 | 62.9 / 76.9 / 86.4 | 323 |
MegaDepth-1500, RoMa v2 matches (5000 per pair), 5 samplings, the same matches for every solver, on one machine: RTX 5060 Ti and a Ryzen 7 5800X (16 threads). Each solver runs at the inlier threshold where it is most accurate. Pairs per second is wall-clock throughput over all 7500 poses: the CPU solvers run one pair per call in 16 single-threaded processes, the GPU solver takes 500 pairs per call. One pair alone takes about 1.7 ms with the kernels, 29 ms with PoseLib and 24 ms with OpenCV RANSAC on one core. The accuracy table above stays on OpenCV RANSAC, as in the papers, so it is comparable with published numbers; this table is the only place the GPU solver is used.
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| <= 0.011 vs the original. GPU, portable bfloat16 and kernels float16 (the defaults): mean |diff| <= 0.05 and p99 <= 0.6 (absolute error of the output keypoints1).
| Check | Tolerance | Measured |
|---|---|---|
| CPU fp32, max |diff| | ≤ 0.011 | not recorded |
| cpu_min_recall | ≤ 0.99 | not recorded |
| cpu_min_match | ≤ 0.99 | not recorded |
| GPU default dtypes, mean |diff| | ≤ 0.05 | 0.00775 |
| GPU default dtypes, p99 |diff| | ≤ 0.6 | 0.40228 |
| gpu_min_recall | ≤ 0.96 | not recorded |
| gpu_min_match | ≤ 0.93 | not recorded |
Recorded for loma-b on RTX 5060 Ti, torch 2.10.0+cu130, 2026-11-17, with these kernels: cudnn_hacccutlass_convigev
Details
LoMa (Nordström, Edstedt et al. 2026): local feature matching with a light transformer. DaD (a VGG11 detector trained by distillation, Edstedt et al. 2025) finds 2048 keypoints per image, a DeDoDe descriptor (VGG for B64/B128, VGG plus DINOv2 for B/L/G/R) describes them, and nine LightGlue-style transformer layers (rotary self-attention, cross-attention) refine the descriptions before a dual-softmax assignment. Returns both keypoint sets, the index pairs of the matches and their scores.
- Input / output
- Input: two RGB images of one scene. Output: for every pixel of the first image its position in the second (x, y in pixels) and a certainty that the pixel is visible in both.
What the numbers mean, and which models need the camera focal length
Sparse matchers (LightGlue) return the keypoints of both images and the index pairs of the matched ones instead. To reconstruct camera poses from an image folder with these matchers, see [structure from motion](../sfm.md).
- Input size
- any H x W >= 32, padded to a multiple of 32; tested up to 3000x4000 (2,424 MiB allocated)
- Options
max_keypoints=2048>= 1· changes the outputlong_side=1024>= 16· changes the outputsize=784>= 56· changes the output
- Outputs
keypoints1,keypoints2,pairs,scores- Checkpoint
dim=256,heads=4,layers=9- CUDA kernels
igevIGEV stereo: convolution epilogues, correlation lookups, GRU gatescutlass_convCUTLASS convolutions with the recurrent update (GRU) fused incudnn_haccconvolutions with fp16 accumulation (cuDNN)
- Apple (MPS)
- fp32
Limitations
- Sparse output: no dense warp. `result.matches()` gives the matched points for `estimate_pose`; `pairs` are the index pairs pycolmap / COLMAP databases take.
- One image pair per call internally (batches run pair by pair); the match counts differ per pair.
- The match count depends on the image pair. The fixed 768 x 1024 LoMa-B64 TensorRT export returns variable-length matches. On ten photos paired with shifted copies, at least 96.7% of TensorRT correspondences agreed with portable within 2 pixels in both images.
- Two resizes per image as upstream: the detector sees a long side of `long_side` (aspect kept), the descriptor a `size` x `size` square whatever the aspect ratio.
- LoMa-R uses rotation augmentation. The other variants are trained for upright images.
Full reference, generated from the repository
LoMa (Nordström, Edstedt et al. 2026): sparse matching of two photos of one scene: DaD keypoints, DeDoDe-G descriptions (VGG19 plus DINOv2) and a light LightGlue-style transformer matcher; matches as keypoint index pairs (the form COLMAP takes). MIT code and weights; DINOv2 Apache-2.0.
Commercial use: yes.
| variant | input | backend | time (ms) | peak memory (MiB, allocated / reserved) |
|---|---|---|---|---|
loma-b | synthetic (768x1024) | original portable kernels portable + torch.compile TensorRT fp16 | 1036 308 – 285 planned for 0.2.1 | 4176 / 4390 1767 / 1878 – 2870 / 3216 – |
loma-b64 | synthetic (768x1024) | original portable kernels portable + torch.compile TensorRT bf16 | – 129 – 101 109 | – 854 / 962 – 1230 / 1356 1010 / 1034 |
loma-b128 | synthetic (768x1024) | original portable kernels portable + torch.compile TensorRT fp16 | – 150 – 122 fails: RuntimeError: Trying to create tensor with negative dimensio | – 878 / 984 – 1398 / 1760 – |
loma-l | synthetic (768x1024) | original portable kernels portable + torch.compile TensorRT fp16 | – 319 – 295 fails: RuntimeError: Trying to create tensor with negative dimensio | – 1903 / 2012 – 3006 / 3194 – |
Details
- Original code: https://github.com/davnords/loma at
ecc947d44ddc - Code licence: MIT AND Apache-2.0 (permissive); files:
LICENSE,LICENSE-DAD,LICENSE-DEDODE,LICENSE-DINOv2 - Weights licence: MIT AND Apache-2.0 (assumed from code)
- Licence evidence: https://github.com/davnords/loma/blob/ecc947d44ddc9d50387e887bbedc743d65ffac00/README.md#license ; https://github.com/Parskatt/dad/blob/41b9e86a1a3ae52128485cc1094aa4952ef489ba/LICENSE ; https://github.com/Parskatt/DeDoDe/blob/6d156183f4dc84cd704ae779eebc8350995c5b06/README.md#license
- Training data: MegaDepth (Internet photo collections with SfM / MVS depth) (Matcher: MegaDepth (paper); DaD: MegaDepth-based distillation (DaD paper); DeDoDe-G descriptor: MegaDepth; B, L, G and R include frozen DINOv2 (LVD-142M). B64 and B128 use VGG descriptors.)
- Options:
max_keypoints= 2048 (>= 1),long_side= 1024 (>= 16),size= 784 (>= 56) - Kernel libraries:
igev,cutlass_conv,cudnn_hacc - Variants:
loma-b,loma-b64,loma-b128,loma-l,loma-g,loma-r - Not in this release (planned for 0.2.1):
loma-bONNX / TensorRT export: the export fails in the network call (negative tensor dimension -1) - Outputs: keypoints1, keypoints2, pairs, scores
- Input size: any H x W >= 32, padded to a multiple of 32
- Tested up to (portable, fp16, RTX 5060 Ti 16 GB):
loma-b3000x4000 (2.4 GiB peak allocated, 3.6 reserved);loma-b643000x4000 (1.1 GiB peak allocated, 1.4 reserved);loma-b1283000x4000 (1.1 GiB peak allocated, 1.4 reserved);loma-l3000x4000 (2.1 GiB peak allocated, 2.8 reserved);loma-g3000x4000 (2.6 GiB peak allocated, 3.3 reserved);loma-r3000x4000 (2.0 GiB peak allocated, 2.7 reserved) loma-b768x1024, first call: portable + torch.compile 104 s (cold Inductor cache).loma-b64768x1024, first call: portable + torch.compile 66 s (cold Inductor cache), TensorRT bf16 102 s (ONNX + engine).loma-b128768x1024, first call: portable + torch.compile 67 s (cold Inductor cache).loma-l768x1024, first call: portable + torch.compile 101 s (cold Inductor cache).loma-boriginal 768x1024: upstream LoMa(LoMaB(mp=False)), float32, 2048 keypointsloma-bload / first call: original 8.52 s / 1.373 s, portable 2.53 s / 1.269 sloma-b64load / first call: portable 0.28 s / 0.759 sloma-b128load / first call: portable 0.52 s / 0.636 sloma-lload / first call: portable 1.49 s / 0.69 s- Protocol, first call, 3 warm-up, peak reset, median of 3 rounds x median of 5
- Speed and memory not measured for:
loma-g,loma-r.
Measured
Dense matchers: 5000 matches per pair sampled from the warp by certainty (upstream's balanced KDE sampling); sparse matchers: all their matches of one call (LightGlue, XFeat: LightGlue's MegaDepth-1500 setting; LoFTR, DeDoDe: their card defaults). OpenCV essential-matrix RANSAC (0.5 px threshold, normalised by the mean focal length); AUC of the larger of the rotation and translation-direction errors. The same 300 pairs for all modes, so float32 and bfloat16 compare pair for pair. Original = the original code named under Details, same checkpoint and protocol.
| variant | dataset | original fp32 | portable | portable fp32 | note |
|---|---|---|---|---|---|
loma-b | MegaDepth-1500, every 5th pair, relative pose AUC@5 / @10 / @20 degrees | 54.5 / 71.6 / 83.7 | 56.2 / 72.6 / 83.9 | ||
loma-b64 | MegaDepth-1500, every 5th pair, relative pose AUC@5 / @10 / @20 degrees | 56.0 / 72.0 / 83.1 |
Measured on portable: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 68b2f345, 2026-11-16; portable32: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks c34edca4, 2026-11-17.
Parity
CPU, float32: max |diff| <= 0.011 vs the original. GPU, portable bfloat16 and kernels float16 (the defaults): mean |diff| <= 0.05 and p99 <= 0.6 (absolute error of the output keypoints1).
| variant | kernels error vs original, mean / p99 | GPU, torch | kernels used |
|---|---|---|---|
loma-b | 0.00775 / 0.40228 | RTX 5060 Ti, torch 2.10.0+cu130 | cudnn_hacc@cudnn91501, cutlass_conv@988138e042cb:sm_120, igev@f075827d68c6:sm_120 |
loma-b64 | 0.00718 / 0.39541 | RTX 5060 Ti, torch 2.10.0+cu130 | cudnn_hacc@cudnn91501, cutlass_conv@988138e042cb:sm_120, igev@f075827d68c6:sm_120 |
loma-b128 | 0.00775 / 0.40228 | RTX 5060 Ti, torch 2.10.0+cu130 | cudnn_hacc@cudnn91501, cutlass_conv@988138e042cb:sm_120, igev@f075827d68c6:sm_120 |
loma-l | 0.00775 / 0.40228 | RTX 5060 Ti, torch 2.10.0+cu130 | cudnn_hacc@cudnn91501, cutlass_conv@988138e042cb:sm_120, igev@f075827d68c6:sm_120 |
loma-g | 0.00775 / 0.40228 | RTX 5060 Ti, torch 2.10.0+cu130 | cudnn_hacc@cudnn91501, cutlass_conv@988138e042cb:sm_120, igev@f075827d68c6:sm_120 |
loma-r | 0.00775 / 0.40228 | RTX 5060 Ti, torch 2.10.0+cu130 | cudnn_hacc@cudnn91501, cutlass_conv@988138e042cb:sm_120, igev@f075827d68c6:sm_120 |
Limitations
- Sparse output: no dense warp.
result.matches()gives the matched points forestimate_pose;pairsare the index pairs pycolmap / COLMAP databases take. - One image pair per call internally (batches run pair by pair); the match counts differ per pair.
- The match count depends on the image pair. The fixed 768 x 1024 LoMa-B64 TensorRT export returns variable-length matches. On ten photos paired with shifted copies, at least 96.7% of TensorRT correspondences agreed with portable within 2 pixels in both images.
- Two resizes per image as upstream: the detector sees a long side of
long_side(aspect kept), the descriptor asizexsizesquare whatever the aspect ratio. - LoMa-R uses rotation augmentation. The other variants are trained for upright images.
Example output
No example rendered for this checkpoint yet.