# Quickshift++: Provably Good Initializations for Sample-Based Mean Shift

arXiv: https://arxiv.org/abs/1805.07909 · HTML: https://mlsanity.com/paper/1805.07909/

## Key facts

- Cited by 14 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 21 May 2018.

## Built on by (most cited first, 14 of 14)

- [MeanShift++: Extremely Fast Mode-Seeking With Applications to Segmentation and Object Tracking](https://mlsanity.com/paper/2104.00303/) (2021)
- [Fast and explainable clustering based on sorting](https://mlsanity.com/paper/2202.01456/) (2022)
- [An Efficient One-Class SVM for Anomaly Detection in the Internet of Things](https://mlsanity.com/paper/2104.11146/) (2021)
- [Git: Clustering Based on Graph of Intensity Topology](https://mlsanity.com/paper/2110.01274/) (2021)
- [SDiT: Semantic Region-Adaptive for Diffusion Transformers](https://mlsanity.com/paper/2601.12283/) (2026)
- [Clustering by Hill-Climbing: Consistency Results](https://mlsanity.com/paper/2202.09023/) (2022)
- [Unsupervised Pre-Training for 3D Leaf Instance Segmentation](https://mlsanity.com/paper/2401.08720/) (2024)
- [Pre-Clustering Point Clouds of Crop Fields Using Scalable Methods](https://mlsanity.com/paper/2107.10950/) (2021)
- [An Asymptotic Equivalence between the Mean-Shift Algorithm and the Cluster Tree](https://mlsanity.com/paper/2111.10298/) (2021)
- [How to scale hyperparameters for quickshift image segmentation](https://mlsanity.com/paper/2201.09286/) (2022)
- [TANGO: Clustering with Typicality-Aware Nonlocal Mode-Seeking and Graph-Cut Optimization](https://mlsanity.com/paper/2408.10084/) (2024)
- [Beyond Attribution: Unified Concept-Level Explanations](https://mlsanity.com/paper/2410.12439/) (2024)
- [Bagged k-Distance for Mode-Based Clustering Using the Probability of Localized Level Sets](https://mlsanity.com/paper/2210.09786/) (2022)
- [On the Burstiness of Faces in Set](https://mlsanity.com/paper/2506.20312/) (2025)

## Abstract (arXiv)

We provide initial seedings to the Quick Shift clustering algorithm, which approximate the locally high-density regions of the data. Such seedings act as more stable and expressive cluster-cores than the singleton modes found by Quick Shift. We establish statistical consistency guarantees for this modification. We then show strong clustering performance on real datasets as well as promising applications to image segmentation.