Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..

Online Segment Any 3D Thing as Instance Tracking

Authors: Hanshi Wang, Cai Zijian, Jin Gao, Yiwei Zhang, Weiming Hu, Ke Wang, Zhipeng Zhang

NeurIPS 2025 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4 Experiments 4.1 Experiment Settings Following our baseline ESAM [6], we begin by training a single-view perception model on Scan Net(200)-25k, a subset of Scan Net200 [7] with RGB-D frames. Then we fine-tune it on RGB-D sequences with full loss functions and randomly sample 8 RGB-D frames per scene at each training step. For the optimization settings, we use an Adam W optimizer with a learning rate of 0.0001 and a weight decay of 0.05 and the batch size is set to 4. All experiments are conducted using Py Torch on a single NVIDIA Tesla A100 GPU. Our experiments are conducted on Scan Net [8], Scan Net200 [7], Scene NN [9], and 3RScan [10] datasets.
Researcher Affiliation Collaboration 1State Key Laboratory of Multimodal Artificial Intelligence Systems (MAIS), CASIA 2School of Artificial Intelligence, University of Chinese Academy of Sciences 3Auto Lab, School of Artificial Intelligence, Shanghai Jiao Tong University 4Anyverse Intelligence 5Beijing Key Laboratory of Super Intelligent Security of Multi-Modal Information 6School of Information Science and Technology, Shanghai Tech University 7Kargo Bot
Pseudocode No The paper describes the methodologies and architectures (LTM, STM, SCL) using descriptive text and mathematical equations, but it does not include any explicitly labeled 'Pseudocode' or 'Algorithm' blocks, nor does it present structured steps in a code-like format.
Open Source Code Yes Code is at https://github.com/Auto Lab-SAI-SJTU/Auto Seg3D.
Open Datasets Yes Our experiments are conducted on Scan Net [8], Scan Net200 [7], Scene NN [9], and 3RScan [10] datasets.
Dataset Splits No The paper mentions training on a subset of Scan Net200 ('Scan Net(200)-25k') and fine-tuning on RGB-D sequences by 'randomly sampling 8 RGB-D frames per scene'. It also describes training on Scan Net and evaluating on Scan Net and Scene NN. However, it does not provide specific train/validation/test split percentages, sample counts for each split, or references to predefined standard splits for reproduction within each dataset.
Hardware Specification Yes All experiments are conducted using Py Torch on a single NVIDIA Tesla A100 GPU.
Software Dependencies No The paper mentions 'Py Torch' but does not specify its version number or any other software dependencies with their respective version numbers.
Experiment Setup Yes For the optimization settings, we use an Adam W optimizer with a learning rate of 0.0001 and a weight decay of 0.05 and the batch size is set to 4.