object-centric-learning

[ICLR 2026 Oral] Latent Particle World Models official repository

Code release for NeurIPS 2023 paper SlotDiffusion: Object-centric Learning with Diffusion Models

Official implementation of the CVPR'24 paper [Adaptive Slot Attention: Object Discovery with Dynamic Slot Number]

Official PyTorch implementation of "Entity-Centric Reinforcement Learning for Object Manipulation from Pixels", Haramati et al., ICLR 2024

Official implementation of: "Object-Centric Video Prediction via Decoupling of Object Dynamics and Interactions" by Villar-Corrales et al. ICIP 2023

Official implementation of: "PlaySlot: Learning Inverse Latent Dynamics for Controllable Object-Centric Video Prediction and Planning" by Villar-Corrales & Behnke. ICML 2025

Official PyTorch implementation of "Hierarchical Entity-centric Reinforcement Learning with Factored Subgoal Diffusion", Haramati et al., ICLR 2026

Official implementation of: "TextOCVP: Object-Centric Video Prediction with Language Guidance" by Villar-Corrales, Plepi & Behnke. TMLR. 2026

This is an implementation of the paper "Are We Done with Object-Centric Learning?"

Official code repository for NeurIPS 2024 paper "Recurrent Complex-Weighted Autoencoders for Unsupervised Object Discovery"

[CVPR 2026] Official PyTorch Implementation of "SlotCurri: Reconstruction-Guided Slot Curriculum: Addressing Object Over-Fragmentation in Video Object-Centric Learning"

🔥unMORE in PyTorch (ICML 2025)

[ECCV2026] Official Pytorch Implementation for "Selective Synergistic learning for Video Object-Centric Learning"

Official code repository for NeurIPS 2023 paper "Contrastive Training of Complex-Valued Autoencoders for Object Discovery"

[ICLR'24] Learning to Compose: Improving Object Centric Learning by Injecting Compositionality

🔥EvObj in PyTorch (CVPR 2026)

Unofficial implementation of SAVi in PyTorch

Object-Centric Disentangled Mechanisms

Constructing Grounded Conceptual Representations via Object-Centric Robotic Active Vision

A world model for underwater ROV salvage. It predicts where the object, the gripper and the surrounding scene move next, from several camera views and the controls. One file, trains and tests end to end on GPU or CPU, no dataset needed.