RT-MonSter++
The real-time version of MonSter++ (Autel Robotics and Huazhong University of Science and Technology), using a small mono depth model. Non-commercial licence.
stable Fork of Junda24/MonSter-plusplus@9167101 · backends: portable, kernels
- Paper MonSter++: Unified Stereo Matching, Multi-view Stereo, and Real-time Stereo with Monodepth Priors arXiv · first released 15 Jan 2025
Checkpoint rt-monster-plusplus-mix-all
Paper 2025 Added in 0.1.0 Evaluated with 0.2.0
Use it
pip install fastforks Lean install with only what this model needs: pip install "fastforks-core[nc]"
from fastforks import stereo
model = stereo.load("rt-monster-plusplus-mix-all") # kernels on a supported GPU, else portable PyTorch
result = model(left, right)
result.disparity
result.backend # what ran: mode, device, dtype, kernels The weights are non-commercial: fastforks downloads them only with FASTFORKS_NONCOMMERCIAL=1 set.
fastforks bench rt-monster-plusplus-mix-all 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 fp32 | fp32 | 252 | 1.00× | 2,922 MiB/3,024 MiB | – | – |
| Original fp16 | fp16 | 251 | – | 2,922 MiB/3,044 MiB | – | – |
| Portable fp32 | fp32 | not measured | – | – | – | – |
| Portable | fp16 | 113 | 2.2× | 2,581 MiB/2,618 MiB | 0.6 | 0.11 |
| Kernels | fp16 | 89.5 | 2.8× | 2,586 MiB/2,640 MiB | 0.9 | 0.09 |
| Portable + torch.compile | 77.0 | 3.3× | 2,750 MiB/2,942 MiB | 69 (build) | – | |
| TensorRT fp16 | fp16 | 44.1 | 5.7× | 952 MiBenginenot comparable | 150 (build) | – |
Measured on RTX 5060 Ti 16 GB, torch 2.14.1+cu130, 2026-11-01 to 2026-11-16. 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 | 917 | 6,269 MiB process | matches | 994x1482 |
| Portable | fp16 | 682 | 7,047 MiB process | matches | 994x1482 | |
| Portable | fp32 | 663 | 6,986 MiB process | matches | 994x1482 | |
| Kernels | fp16 | not implemented | – | 994x1482 | ||
| NVIDIA L4 CUDA, torch 2.14.1+cu130, 2026-11-08 | Portable | fp16 | 132 | 2,565 MiB/2,618 MiB | – | 994x1482 |
| Kernels | fp16 | 125 | 2,646 MiB/3,100 MiB | – | 994x1482 | |
| NVIDIA L40S CUDA, torch 2.14.1+cu130, 2026-11-07 | Portable | fp16 | 48.3 | 2,565 MiB/2,618 MiB | – | 994x1482 |
| Kernels | fp16 | 50.1 | 2,646 MiB/3,100 MiB | – | 994x1482 | |
| NVIDIA A10 CUDA, torch 2.14.1+cu130, 2026-11-08 | Portable | fp16 | 112 | 2,565 MiB/2,618 MiB | – | 994x1482 |
| Kernels | fp16 | 105 | 2,646 MiB/3,120 MiB | – | 994x1482 | |
| NVIDIA A100-SXM4-40GB CUDA, torch 2.14.1+cu130, 2026-11-07 | Portable | fp16 | 49.4 | 2,565 MiB/2,618 MiB | – | 994x1482 |
| Kernels | fp16 | 54.3 | 2,646 MiB/3,040 MiB | – | 994x1482 | |
| NVIDIA RTX PRO 6000 CUDA, torch 2.14.1+cu130, 2026-11-08 | Portable | fp16 | 25.3 | 2,589 MiB/2,658 MiB | – | 994x1482 |
| Kernels | fp16 | 23.6 | 2,641 MiB/3,080 MiB | – | 994x1482 | |
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
Non-commercial only
- Code
- LicenseRef-noncommercial LICENSE
- Weights
- LicenseRef-noncommercial source
- Training data
- mix_all (Zero_shot.pth): 'all the datasets we collect over 2M image pairs' (README), not listed
- code: LicenseRef-noncommercial (noncommercial)
- weights: LicenseRef-noncommercial
Our reading of the licences, not legal advice.
Weights
Zero_shot.pth 172 MB· hosted on Hugging Facefastforks downloads and checks these files on first use (fastforks download rt-monster-plusplus-mix-all). Non-commercial weights need FASTFORKS_NONCOMMERCIAL=1.
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.
ETH3D two-view train
bad > 1 px % / EPEtraining data not stated: possibly seen
| Run | dtype | bad px % ↓ | EPE ↓ | Δ bad px % vs fp32 original | pairs | ms / pair |
|---|---|---|---|---|---|---|
| Original fp32 | fp32 | 2.05 | 0.226 | reference | 27 | 65.4 |
| Original fp16 | fp16 | 2.05 | 0.226 | ±0.00 | 27 | 62.2 |
| Portable fp32 | fp32 | 2.05 | 0.226 | ±0.00 | 27 | 49.0 |
| Portable | fp16 | 2.05 | 0.226 | ±0.00 | 27 | 43.6 |
| Kernels | fp16 | 2.05 | 0.226 | ±0.00 | 27 | 36.3 |
Original = the authors’ code, same checkpoint and protocol. Measured on original fp32, original fp16: RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-05; portable fp32, portable: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 468fba10, 2026-11-05; kernels: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 939adfe8, 2026-11-06.
Middlebury 2014 trainingH
bad > 2 px % / EPEtraining data not stated: possibly seen
| Run | dtype | bad px % ↓ | EPE ↓ | Δ bad px % vs fp32 original | pairs | ms / pair |
|---|---|---|---|---|---|---|
| Original fp32 | fp32 | 11.01 | 1.988 | reference | 15 | 215 |
| Original fp16 | fp16 | 11.01 | 1.988 | ±0.00 | 15 | 212 |
| Portable fp32 | fp32 | 11.01 | 1.989 | ±0.00 | 15 | 174 |
| Portable | fp16 | 11.00 | 1.989 | +0.01 | 15 | 116 |
| Kernels | fp16 | 11.02 | 1.998 | −0.01 | 15 | 88.1 |
Original = the authors’ code, same checkpoint and protocol. Measured on original fp32, original fp16: RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-05; portable fp32, portable: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 468fba10, 2026-11-05; kernels: RTX 5060 Ti, torch 2.14.1+cu130, fastforks 939adfe8, 2026-11-06.
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| <= 0.002 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.02 px and p99 <= 0.25 px.
| Check | Tolerance | Measured |
|---|---|---|
| CPU fp32, max |diff| | ≤ 0.002 px | not recorded |
| GPU default dtypes, mean |diff| | ≤ 0.02 px | 0.0032 px |
| GPU default dtypes, p99 |diff| | ≤ 0.25 px | 0.01047 px |
Recorded for rt-monster-plusplus-mix-all on RTX 5060 Ti, torch 2.14.1+cu130, 2026-10-31, with these kernels: cudnn_hacccutlass_convfastbbigev
Details
RT-MonSter++ (Cheng et al. 2025), the real-time MonSter++: Depth Anything V2 (ViT-S) monocular branch + coarse-to-fine (1/16, 1/8, 1/4) local geometry volumes, stereo and mono ConvGRUs, 4 updates.
- Input / output
- Input: a rectified left / right image pair. Output: a disparity map (pixels, left view).
- Input size
- any H x W >= 32, padded to a multiple of 32; tested up to 1988x2880 (9,486 MiB allocated)
- Options
cuda_graphs=true
- Outputs
disparity- CUDA kernels
igevIGEV stereo: convolution epilogues, correlation lookups, GRU gatescutlass_convCUTLASS convolutions with the recurrent update (GRU) fused incudnn_haccconvolutions with fp16 accumulation (cuDNN)fastbbfast backbone kernels: flash attention, window attention, instance norm and rotary embeddings for ViT and similar backbones
- Apple (MPS)
- fp32, fp16
Limitations
- Kernels at 994 x 1482 peak 2631 MB (reserved 3436) vs 2581 MB (2658) portable and 2922 MB original, 21 % faster than portable; the difference is cached CUDA-graph / cuDNN workspace blocks.
Full reference, generated from the repository
The real-time version of MonSter++ (Autel Robotics and Huazhong University of Science and Technology), using a small mono depth model. Non-commercial licence.
Commercial use: no (code: LicenseRef-noncommercial (noncommercial); weights: LicenseRef-noncommercial).
| variant | input | backend | time (ms) | peak memory (MiB, allocated / reserved) |
|---|---|---|---|---|
rt-monster-plusplus-mix-all | Middlebury 2014 half (994x1482) | original (fp32) original portable kernels portable + torch.compile TensorRT fp16 | 252 251 113 89.5 77.0 44.1 | 2922 / 3024 2922 / 3044 2581 / 2618 2586 / 2640 2750 / 2942 1139 / 1210 |
Details
- Original code: https://github.com/Junda24/MonSter-plusplus at
9167101af8fb - Code licence: LicenseRef-noncommercial (noncommercial); files:
LICENSE - Weights licence: LicenseRef-noncommercial
- Licence evidence: https://github.com/Junda24/MonSter-plusplus/blob/9167101af8fbe93c706707daa7a8181089c8b1b4/LICENSE#L34
- Training data: mix_all (Zero_shot.pth): 'all the datasets we collect over 2M image pairs' (README), not listed
- Options:
cuda_graphs= True (speed only) - Kernel libraries:
igev,cutlass_conv,cudnn_hacc,fastbb - Variants:
rt-monster-plusplus-mix-all - Outputs: disparity
- Input size: any H x W >= 32, padded to a multiple of 32
- Tested up to (portable, fp16, RTX 5060 Ti 16 GB):
rt-monster-plusplus-mix-all1988x2880 (9.3 GiB peak allocated, 11.1 reserved) rt-monster-plusplus-mix-all994x1482, first call: portable + torch.compile 69 s (cold Inductor cache), TensorRT fp16 150 s (ONNX + engine).rt-monster-plusplus-mix-allload / first call: original 0.99 s / 0.662 s (375x1242), portable 0.63 s / 0.389 s (375x1242), kernels 0.93 s / 0.352 s (375x1242)- Protocol, first call, 3 warm-up, peak reset, median of 3 rounds x median of 5
Measured
Original = the original code named under Details, same checkpoint and protocol.
| variant | dataset | original fp32 | original fp16 | portable | portable fp32 | kernels | note |
|---|---|---|---|---|---|---|---|
rt-monster-plusplus-mix-all | ETH3D two-view train, bad > 1 px % / EPE | 2.05 / 0.226 | 2.05 / 0.226 | 2.05 / 0.226 | 2.05 / 0.226 | 2.05 / 0.226 | possibly in the training data |
rt-monster-plusplus-mix-all | Middlebury 2014 trainingH, bad > 2 px % / EPE | 11.01 / 1.988 | 11.01 / 1.988 | 11.00 / 1.989 | 11.01 / 1.989 | 11.02 / 1.998 | possibly in the training data |
Measured on portable: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 468fba10, 2026-11-05; portable32: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 468fba10, 2026-11-05; kernels: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, fastforks 939adfe8, 2026-11-06.
Parity
CPU, float32: max |diff| <= 0.002 vs the original. GPU, float16 (the default), portable and kernels: mean |diff| <= 0.02 px and p99 <= 0.25 px.
| variant | kernels error vs original, mean / p99 | GPU, torch | kernels used |
|---|---|---|---|
rt-monster-plusplus-mix-all | 0.0032 / 0.01047 | RTX 5060 Ti, torch 2.14.1+cu130 | cudnn_hacc@cudnn92400, cutlass_conv@988138e042cb:sm_120, fastbb@7d058322634d:sm_120, igev@f075827d68c6:sm_120 |
Limitations
- Kernels at 994 x 1482 peak 2631 MB (reserved 3436) vs 2581 MB (2658) portable and 2922 MB original, 21 % faster than portable; the difference is cached CUDA-graph / cuDNN workspace blocks.
Example output
One fixed input through this checkpoint, the same for every model of the task (kernels · 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).