OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Monoscene

MonoScene is an open-source computer-vision research implementation that predicts 3D semantic occupancy from a single image. Its Python and PyTorch repository documents dataset preparation, model inference and output visualization for NYUv2, SemanticKITTI and KITTI-360.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Monoscene — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use MonoScene when you need a research-code workflow for single-image semantic 3D occupancy prediction and can work with its documented Python, PyTorch and CUDA versions and dataset requirements.

Who it is for

  • Computer-vision researchers studying monocular 3D scene completion or occupancy prediction.
  • Developers prototyping image-based 3D perception workflows.
  • Teams evaluating a research implementation on the datasets named in its documentation.

Documented uses

  • Generate semantic 3D occupancy predictions from images using the documented model workflow.
  • Prepare NYUv2 or SemanticKITTI data for the repository's workflow.
  • Run documented inference on NYUv2, SemanticKITTI or KITTI-360.
  • Visualize prediction outputs using the documented visualization approach.
Before you choose

Requirements & limits

The code repository is public and specifies the Apache-2.0 license. Its documented setup is a local Python environment with PyTorch and CUDA.

  • The documented software stack is Python 3.7, PyTorch 1.7.1 and CUDA 10.2, plus additional dependencies.
  • Dataset preparation and path configuration are part of the workflow; NYUv2 and SemanticKITTI require preprocessing.
  • The repository documents KITTI-360 for inference, not training.
  • The documented visualization uses Mayavi and points to Open3D-based code as an alternative.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Generate predictions for a supported dataset

Prepare a documented dataset and run MonoScene's inference script to write predictions to an output folder.

Before you start

  • A MonoScene checkout and an environment using the documented Python 3.7, PyTorch 1.7.1 and CUDA 10.2 stack, with the listed dependencies installed.
  • A supported dataset: NYUv2, SemanticKITTI or KITTI-360. Dataset paths and preprocessing data must be prepared as described for the chosen dataset.
  • The corresponding pretrained model placed in the repository's trained_models folder.

Steps

  1. Follow the repository's installation instructions and prepare the chosen dataset; run the documented preprocessing steps for NYUv2 or SemanticKITTI.

  2. Create an output folder and set MONOSCENE_OUTPUT to its path; set the dataset-specific root and preprocessing path variables required by the inference command.

  3. From the repository, run monoscene/scripts/generate_output.py with the documented dataset value and paths: NYU, kitti or kitti_360. For KITTI-360, provide a sequence as well.

Expected outcome

The inference script is configured to generate predictions in the specified output folder for the selected dataset.

Guide limitations

  • The documented inference targets are the NYUv2 test set, SemanticKITTI validation set and KITTI-360 sequences.
  • KITTI-360 is documented for inference only.
  • The guide describes repository instructions, not a verified execution or a guarantee of compatibility with other environments.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Model inference and serving

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
MonosceneGenerate semantic 3D occupancy predictions from images using the documented model workflow.CloudThe code repository is public and specifies the Apache-2.0 license. Its documented setup is a local Python environment with PyTorch and CUDA.
Pulse Of MotionEstimate a video's physical frame rate from its visual dynamics.LinuxThe repository is public. Its README describes a released checkpoint that downloads from Hugging Face on first inference, with a manual download option. The strongest internal model is not fully released; the README directs interested users to contact the project about access. No license information is provided in the repository evidence.
Neural Image AssessmentTrain the model using AVA images and rating annotations.Cloud, Linux, macOS, Windows
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
astra-vision/MonoScene · Official homepage title · retained in its source language
How we review sourcesSuggest a correction