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

Multinet

MultiNet is an open-source computer-vision model that jointly performs road segmentation, car detection and street classification for autonomous-driving scenes. It combines task-specific TensorFlow models in an encoder-decoder architecture with a VGG encoder and independent decoders.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Multinet — 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 MultiNet when you need a codebase that combines road segmentation, car detection and street classification in one autonomous-driving perception model, and can work with its documented legacy Python and TensorFlow requirements.

Who it is for

  • Computer-vision developers exploring combined road-scene perception tasks.
  • Researchers or practitioners who want to run the included image demo or evaluate a trained model.
  • Developers adapting the documented training configuration to their own data.

Documented uses

  • Generate a prediction for a road-scene image using the included demo script.
  • Evaluate a trained model with the repository’s evaluation script.
  • Train the documented MultiNet2 configuration or adapt MultiNet3 for joint detection and segmentation.
  • Organize experiment data and run outputs in configurable directories.
Before you choose

Requirements & limits

The repository is public and specifies the MIT license.

  • The README specifies Python 2.7 and TensorFlow 1.0, alongside six Python libraries; these version requirements should be considered before adopting the code.
  • Demo setup requires initializing submodules and building Cython utilities. KITTI Road data and the optional evaluation-code build are described for training workflows.
  • MultiNet3 training supports joint detection and segmentation out of the box; classification training data is not public, while the README describes full-model training with user-provided data.
  • The model combines KittiSeg, KittiBox and KittiClass components and is designed to work with the TensorVision backend.

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

Set up MultiNet for the included image demo

Prepare the repository and its Cython utilities for the documented demo workflow.

Before you start

  • Python 2.7 and TensorFlow 1.0, plus matplotlib, NumPy, Pillow, SciPy, runcython and commentjson.
  • A local environment with Git and the ability to build the Cython utility code.

Steps

  1. Clone the MultiNet repository.

  2. Initialize all submodules with the command below.

    Documented command
    git submodule update --init --recursive

  3. Run make from submodules/KittiBox/submodules/utils/ to build the Cython code.

  4. Use the README’s demo command with the included image: the command below.

    Documented command
    python demo.py --gpus 0 --input data/demo/um_000005.png

Expected outcome

The README describes obtaining a prediction for the included sample image.

Guide limitations

  • This guide reflects the README’s Python 2.7 and TensorFlow 1.0 requirements; no compatibility guarantee is implied.
  • The setup steps are documented instructions, not a claim that the demo was executed.
  • KITTI Road data is not required for the demo, but is listed for training.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
MarvinTeichmann/MultiNet · Official homepage title · retained in its source language
How we review sourcesSuggest a correction