YOLOv18 – Repository Overview and Current Project Information

YOLOv18 is the name used by a public GitHub project maintained under the FrancescoSaverioZuppichini account.

The repository identifies itself as:

Official YOLOv18

and provides a Python package installation command for yolov18.

At its current stage, however, the repository is extremely lightweight and does not contain a complete object detection implementation.

What Is YOLOv18?

YOLOv18 is currently presented as a Python project associated with the YOLO object detection family.

The project README provides:

pip install yolov18

as its installation method.

The repository also refers to an implementation of a YOLOv18 paper, but the linked item currently points to YouTube rather than a detailed technical paper page.

Because the repository does not yet expose a full model implementation, specific architectural claims cannot currently be verified.

YOLOv18 Repository Status

The current repository contains only a small number of files and directories:

  • .github/workflows
  • docs
  • Makefile
  • README.md
  • requirements.txt
  • setup.py

The repository currently shows only 1 commit.

There are also currently:

  • 0 forks
  • 0 stars
  • 0 open issues
  • 0 pull requests

visible on the repository page.

This suggests that the project is still at a very early or placeholder stage.

YOLOv18 Installation

The current README provides a simple installation command:

pip install yolov18

This suggests that YOLOv18 is intended to be distributed as a Python package.

However, the repository does not currently document how to use the installed package for object detection.

YOLOv18 Architecture

The current repository does not provide enough source code to verify a specific YOLOv18 architecture.

There are no visible model implementation directories documenting:

  • Backbone
  • Neck
  • Detection head
  • Attention modules
  • Transformer blocks
  • Convolution blocks
  • Anchor configuration
  • Bounding box regression
  • Feature pyramid structure

Therefore, any detailed YOLOv18 architecture description would currently be speculative.

YOLOv18 Backbone

No verified backbone is currently documented.

It is not possible to confirm whether YOLOv18 uses:

  • CSP-style blocks
  • C2f
  • ELAN
  • GELAN
  • Transformers
  • Attention
  • Hypergraphs
  • Depthwise separable convolutions

based on the available repository.

YOLOv18 Detection Head

The repository does not currently document the detection head.

As a result, it is not possible to verify whether YOLOv18 uses:

  • Anchor-based detection
  • Anchor-free detection
  • Decoupled heads
  • End-to-end prediction
  • NMS-free detection

These should not be treated as confirmed YOLOv18 features.

YOLOv18 Model Variants

The repository does not currently list model variants such as:

  • YOLOv18n
  • YOLOv18s
  • YOLOv18m
  • YOLOv18l
  • YOLOv18x

No model-size comparison table is available in the current repository.

YOLOv18 Pretrained Weights

The repository currently shows no published releases or pretrained model checkpoints.

There are no visible weights for any YOLOv18 model variant.

YOLOv18 and COCO

The README includes the statement:

We achieved skynet on COCO dataset

but it does not provide quantitative benchmark evidence alongside that claim.

There are currently no published values for:

  • mAP50
  • mAP50-95
  • AP50
  • AP75
  • Precision
  • Recall
  • Parameters
  • FLOPs
  • FPS
  • GPU latency
  • CPU latency

Therefore, the COCO statement should not be treated as a verified benchmark result.

YOLOv18 Training

The current repository does not provide documented training instructions.

There are no visible examples explaining:

  • Dataset preparation
  • COCO training
  • Custom dataset training
  • Epoch count
  • Image size
  • Batch size
  • Optimizer
  • Learning rate
  • Data augmentation
  • Multi-GPU training
  • Transfer learning

A reliable YOLOv18 training tutorial therefore cannot currently be created from this repository alone.

YOLOv18 Custom Dataset Training

Custom dataset support is not currently documented.

A typical YOLO training setup would require:

  • Training images
  • Validation images
  • Bounding box labels
  • Class names
  • Dataset configuration
  • Model configuration

But these components are not currently described by the repository.

YOLOv18 Inference

There is no documented inference workflow in the current repository.

The project does not currently provide visible examples for:

  • Loading a YOLOv18 model
  • Loading pretrained weights
  • Predicting an image
  • Processing video
  • Using a webcam
  • Processing streams
  • Batch inference

Prediction behavior therefore cannot yet be documented accurately.

YOLOv18 Validation

The repository does not provide a documented validation pipeline.

There are no visible examples for calculating:

  • Precision
  • Recall
  • mAP50
  • mAP50-95
  • COCO AP

Without model code and weights, performance cannot currently be reproduced.

YOLOv18 Export

The repository does not currently document export support.

There is no verified information showing support for:

  • ONNX
  • TensorRT
  • OpenVINO
  • CoreML
  • TensorFlow
  • TensorFlow Lite

These should not be presented as YOLOv18 capabilities until implementation is provided.

YOLOv18 Performance

No verified performance table is currently available.

MetricCurrent YOLOv18 Repository
mAP50-95Not provided
mAP50Not provided
ParametersNot provided
FLOPsNot provided
FPSNot provided
GPU latencyNot provided
CPU latencyNot provided
Model variantsNot provided

Direct performance comparison with other YOLO generations is therefore not currently possible.

YOLOv18 vs YOLOv16

The current YOLOv18 repository does not provide enough technical data to compare it reliably with YOLOv16.

There are no verified figures for:

  • Accuracy
  • Speed
  • Parameters
  • FLOPs
  • Architecture
  • Training efficiency

Any claim that YOLOv18 is faster or more accurate would therefore be unsupported.

YOLOv18 vs YOLOv13

YOLOv13 has a documented model architecture, benchmark results, and implementation details.

The current YOLOv18 repository does not yet contain comparable technical information.

This means a meaningful YOLOv13 vs YOLOv18 benchmark comparison cannot currently be made.

YOLOv18 Repository Structure

The repository currently uses a lightweight project structure.

README.md

The README identifies the project as “Official YOLOv18,” includes a COCO statement, and provides the installation command.

setup.py

The repository includes a Python setup file intended for package distribution.

requirements.txt

A Python requirements file is included.

Makefile

The repository includes a Makefile for project-related commands.

docs

A documentation directory is present, although detailed model documentation is not exposed from the main repository page.

YOLOv18 Development Stage

Based on the visible repository information, YOLOv18 appears to be at a very early stage.

The repository currently has:

  • 1 commit
  • A very short README
  • No visible full model implementation
  • No pretrained weights
  • No benchmark table
  • No documented training pipeline
  • No documented inference pipeline
  • No releases

This is important when evaluating the project.

Is YOLOv18 Production Ready?

There is currently not enough evidence to describe this YOLOv18 project as production-ready.

A mature object detection framework would normally provide:

  • Stable source code
  • Architecture definitions
  • Pretrained weights
  • Training scripts
  • Validation tools
  • Inference examples
  • Reproducible benchmarks
  • Export support
  • Deployment documentation

These are not currently available in the visible repository.

Does YOLOv18 Use PyTorch?

The repository currently does not provide enough information to verify which deep learning framework a full YOLOv18 detector would use.

Python packaging alone does not prove that the model is based on PyTorch.

Is YOLOv18 Anchor-Free?

The current repository does not provide enough information to determine whether YOLOv18 is:

  • Anchor-based
  • Anchor-free

This remains unverified.

Is YOLOv18 NMS-Free?

No inference or post-processing pipeline is currently documented.

It is therefore not possible to determine whether YOLOv18 uses:

  • Non-Maximum Suppression
  • NMS-free end-to-end prediction

Does YOLOv18 Support Segmentation?

Instance segmentation is not currently documented.

Does YOLOv18 Support Classification?

Image classification models are not currently documented.

Does YOLOv18 Support Pose Estimation?

Pose estimation is not currently documented.

Does YOLOv18 Support Oriented Bounding Boxes?

OBB detection is not currently documented.

Advantages of the Current YOLOv18 Project

Simple Installation

The README provides a straightforward Python package installation command:

pip install yolov18

Python Project Structure

The repository contains standard Python packaging files.

Lightweight Repository

The current project is small and easy to inspect.

Current Limitations

The biggest limitation is the lack of an actual publicly visible complete detector implementation.

Currently missing are:

  • Network architecture
  • Backbone
  • Neck
  • Detection head
  • Pretrained weights
  • Model variants
  • Training code
  • Validation code
  • Prediction examples
  • Benchmark results
  • Parameters
  • FLOPs
  • Latency measurements
  • Export workflows
  • Deployment documentation

Frequently Asked Questions

What is YOLOv18?

YOLOv18 is the name used by a public GitHub repository maintained under the FrancescoSaverioZuppichini account. Its README identifies the project as Official YOLOv18.

Who maintains the YOLOv18 repository?

The repository is maintained under the FrancescoSaverioZuppichini GitHub account.

How can YOLOv18 be installed?

The repository currently provides:

pip install yolov18

How many commits does the repository have?

The current repository shows 1 commit.

Does YOLOv18 have pretrained weights?

No visible pretrained weights or releases are currently provided.

Does YOLOv18 have COCO benchmarks?

The README mentions COCO but does not provide standard numerical benchmark results.

What architecture does YOLOv18 use?

The current repository does not contain enough implementation information to verify the architecture.

What YOLOv18 models are available?

No Nano, Small, Medium, Large, or Extra-Large model variants are currently documented.

Is YOLOv18 implemented in PyTorch?

The current repository does not provide enough evidence to verify the deep learning framework.

Can YOLOv18 be trained on custom datasets?

A custom dataset training workflow is not currently documented.

Does YOLOv18 support ONNX?

ONNX export is not currently documented.

Does YOLOv18 support TensorRT?

TensorRT export is not currently documented.

Does YOLOv18 support segmentation?

Segmentation is not currently documented.

Does YOLOv18 support pose estimation?

Pose estimation is not currently documented.

Is YOLOv18 production ready?

The current repository lacks enough implementation, benchmarking, training, and deployment information to establish production readiness.

Conclusion

The current YOLOv18 repository presents itself as an Official YOLOv18 project and provides the installation command:

pip install yolov18

However, the repository currently contains only one commit and a lightweight project structure. It does not expose a complete object detection implementation, pretrained weights, model variants, benchmark results, training instructions, inference examples, or deployment documentation.

For that reason, technical claims about YOLOv18 architecture, accuracy, speed, model size, or improvements over previous YOLO generations should not be made until reproducible implementation and benchmark information becomes available.

Scroll to Top