README.md

metadata

library_name: fastforks
pipeline_tag: depth-estimation
license: other
license_name: noncommercial
license_link: >-
  https://github.com/Junda24/MonSter-plusplus/blob/9167101af8fbe93c706707daa7a8181089c8b1b4/LICENSE#L34
tags:
  - stereo
  - stereo-matching
  - disparity
  - cuda
  - pytorch
  - fastforks

RT-MonSter++ (fastforks)

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.

This repo is the model card only: it holds no weights. fastforks downloads the authors' weights from the links below and checks them against the SHA-256 in the model card. Code: https://github.com/kleinicke/fastforks. Kernels for the kernels backend: https://huggingface.co/fastforks/kernels. Original code: https://github.com/Junda24/MonSter-plusplus at 9167101af8fb.

Author weight repos: https://huggingface.co/cjd24/MonSter-plusplus.

Task: stereo. Tier: experimental. Outputs: disparity. Backends: portable, kernels.

Input: a rectified left / right image pair. Output: a disparity map (pixels, left view).

Variants

Weights are the authors' files (not hosted here). SHA-256 shortened to 12 hex digits; the full hashes are in card.toml.

variant	weights (author link)	SHA-256	weights licence	commercial use
rt-monster-plusplus-mix-all@1	Zero_shot.pth	0741dd6f622b	LicenseRef-noncommercial	no

Licence

Original-code strings, recorded as they are (research-use package; check them before any other use). Code: LicenseRef-noncommercial. Files: LICENSE.

Training data: mix_all (Zero_shot.pth): 'all the datasets we collect over 2M image pairs' (README), not listed.

Licence evidence: https://github.com/Junda24/MonSter-plusplus/blob/9167101af8fbe93c706707daa7a8181089c8b1b4/LICENSE#L34

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-all 1988x2880 (9.3 GiB peak allocated, 11.1 reserved)

Speed and memory

Stereo, Middlebury 2014 half, 994x1482

model	time (ms): original fp32	original	portable	kernels	kernels speed-up: vs original	vs portable	peak memory (MiB, allocated / reserved): original	portable	kernels
rt-monster-plusplus-mix-all	252	251	113	94.1	2.7x	1.2x	2922 / 3044	2592 / 2646	2704 / 3976

Startup and first call

model	load (s): original	portable	kernels	first call (s): original	portable	kernels
rt-monster-plusplus-mix-all	0.99 (375x1242)	0.26	0.23†	0.66 (375x1242)	0.13†	0.10†

torch.compile and TensorRT

model	input	portable ms	compile ms	compile s (cold)	TensorRT fp16 ms	ONNX + build s	kernels ms	peak MiB (alloc / reserved): portable	compile	TensorRT (device)
rt-monster-plusplus-mix-all	994x1482	113	77.0	69.4	44.1	150	94.1	2592 / 2646	2750 / 2942	952

Fixed-size speed and peak memory, all models

model	mode	480x640	720x1280	1080x1920	12 MP 3000x4000
rt-monster-plusplus-mix-all	portable	31.2 ms / 609 / 720	67.1 ms / 1625 / 1968	168.0 ms / 3478 / 4148	OOM
rt-monster-plusplus-mix-all	kernels	18.3 ms / 630 / 1934	57.5 ms / 1660 / 1998	140.1 ms / 3545 / 4188	OOM

Fixed-size cells: time (ms) / peak allocated / peak reserved memory (MiB).

Hardware, protocol and caveats: benchmarks (RTX 5060 Ti 16 GB, sm_120). Original = the authors' code at the precision of portable (fp16, or (fp32) where fp16 is unsupported or not finite); original fp32 = its default fp32 run. Peak memory is the torch allocator's.

Accuracy

On public datasets

Original = the authors' code, same checkpoint and protocol.

variant	dataset	original fp32	original fp16	portable	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	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	11.02 / 1.998	possibly in the training data

Measured on portable: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-05; kernels: NVIDIA GeForce RTX 5060 Ti, torch 2.14.1+cu130, 2026-11-06.

Parity with the original code

Every variant is checked against reference outputs of the original code at its pinned commit. 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.

Usage

from fastforks import stereo

# non-commercial weights: run with FASTFORKS_NONCOMMERCIAL=1 in the environment
model = stereo.load("rt-monster-plusplus-mix-all")  # mode="auto": the preferred backend from recorded measurements
result = model("left.png", "right.png")
result.disparity  # [H, W] float32 tensor, input resolution
result.backend  # what ran, e.g. Backend(mode='portable', device='cuda:0', dtype='float16')

Install with pip install fastforks (see https://github.com/kleinicke/fastforks). Other variants: rt-monster-plusplus-mix-all.

Kernels backend

stereo.load("rt-monster-plusplus-mix-all", mode="kernels") runs our CUDA kernels (igev, cutlass_conv, cudnn_hacc, fastbb). It needs a prebuilt kernel for your GPU architecture (downloaded from https://huggingface.co/fastforks/kernels on first use) and a recent NVIDIA driver; if there is none it raises BackendUnavailableError with the reason and never falls back silently. mode="auto" picks kernels when it can run and logs the choice.