OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Yolo Patch Based Inference

Yolo Patch Based Inference is a Python library for running supported Ultralytics object-detection and instance-segmentation models on image patches, then combining crop-level predictions and suppressing duplicates.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Yolo Patch Based Inference — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when a Python image-analysis workflow needs patch-based inference with a supported Ultralytics model, especially when targeting small objects or combining predictions across overlapping crops.

Who it is for

  • Python developers building computer-vision workflows with Ultralytics models.
  • Teams seeking patch-based inference for small objects in images.
  • Practitioners who need bounding-box detection, instance-segmentation polygons or customized inference-result visualizations.

Documented uses

  • Run a YOLO model on crops from an image.
  • Combine predictions from overlapping crops and suppress duplicates.
  • Retrieve confidence scores, boxes and class names or IDs, and segmentation polygons when available.
  • Customize visualization for direct or patch-based inference results.
Before you choose

Requirements & limits

The README documents pip installation. The public GitHub repository identifies the project as AGPL-3.0 licensed.

  • The documented input image format is BGR, represented as a NumPy array.
  • Crop dimensions, overlap and inference thresholds are set through MakeCropsDetectThem; CombineDetections handles consolidation and duplicate suppression.
  • Supported model examples include YOLOv8, YOLOv9, YOLOv10, YOLO11, YOLO12, FastSAM and RTDETR, including listed segmentation variants.
  • For CUDA use, the README recommends pre-installing PyTorch with CUDA support; it describes CPU installation as the default otherwise.
  • The AGPL-3.0 license should be considered when assessing project fit.

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

Run patch-based inference on an image

Process an image with a supported YOLO model on overlapping crops and combine the resulting detections.

Before you start

  • Python and the patched_yolo_infer package installed with pip.
  • An image supplied as a BGR NumPy array.
  • A supported model file, such as the README example yolo11m.pt.

Steps

  1. Create a MakeCropsDetectThem instance with the image, model path, detection or segmentation setting, crop dimensions, overlap and inference thresholds.

  2. Pass the crop-inference object to CombineDetections and set the non-maximum-suppression threshold.

  3. Read the combined result's image, filtered confidences, boxes and class labels; use polygon results when available for segmentation.

Expected outcome

A combined inference result with object locations, confidence scores and class information, plus polygon coordinates when available.

Guide limitations

  • The README documents the sequence and a code example but does not establish model accuracy or suitability for a particular dataset.
  • When CUDA is available, the README recommends installing PyTorch with CUDA support before the package; otherwise it says the CPU version is installed.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
Koldim2001/YOLO-Patch-Based-Inference · Official homepage title · retained in its source language
How we review sourcesSuggest a correction