OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Image & Video AI

Deepsim

Deepsim is a PyTorch research implementation for learning image-shape manipulation from a single real image paired with a primitive representation, then generating corresponding image changes from an edited primitive.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Deepsim — OpenViST project profile in Image & Video AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Deepsim when you want to investigate single-pair image training and can prepare a matching image and primitive input for local experimentation.

Who it is for

  • Computer-vision researchers studying image-to-image translation from a single example.
  • Developers exploring shape-controlled image generation with edge maps, segmentation, or manual primitives.

Documented uses

  • Train a model to map one image’s primitive representation to the image itself.
  • Edit an edge, segmentation, or combined primitive and generate a corresponding image.
  • Explore image changes such as altering facial features or changing object shapes.
Before you choose

Requirements & limits

  • The repository documents Python 3.7 setup and a requirements-file installation workflow.
  • Primitive choices include edges, segmentation, both together, and manual inputs such as a painting.
  • Segmentation test inputs must match training-input colors exactly.
  • The project is presented as a code implementation with training and testing scripts, rather than a hosted editing interface.

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 the DeepSIM Python environment

Clone the repository and install its listed dependencies in a Python 3.7 virtual environment.

Before you start

  • Python 3.7
  • Access to the DeepSIM repository

Steps

  1. Clone the repository with the command below and enter the DeepSIM directory.

    Documented command
    git clone https://github.com/eliahuhorwitz/DeepSIM.git

  2. Create and activate an environment with the command below and source deepsim_venv/bin/activate.

    Documented command
    python3.7 -m venv deepsim_venv

  3. Install the listed libraries with the command below.

    Documented command
    pip install -r requirements.txt

Expected outcome

A virtual environment with the repository's listed dependencies installed.

Guide limitations

  • These are repository setup instructions; they do not establish that installation or execution has been independently tested.
Documentation 1 ↗
Source-based practical guide

Prepare a training run with a documented primitive

Configure a training command around a dataset and one of DeepSIM's supported primitive types.

Before you start

  • The repository and its Python environment
  • A suitable input file under /train_
  • A primitive choice: seg, edges, seg_edges, or manual

Steps

  1. Choose the primitive type and provide its corresponding training input in the documented dataset folder structure.

  2. Set --dataroot to the dataset location and --primitive to the selected type.

  3. Use a documented example command as a template, such as the command below.

    Documented command
    python3.7 ./train.py --dataroot ./datasets/car --primitive seg --no_instance --tps_aug 1 --name DeepSIMCar

Expected outcome

A configured training command for a single-example image-manipulation experiment.

Guide limitations

  • For segmentation primitives, the test-time input must match the training input exactly in color.
  • For automatic edges, the repository says the train_edges folder may be left empty so an edge map is generated.
  • The example commands do not guarantee a successful training run.
Documentation 1 ↗
Evidence and transparency

Sources & review

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