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..
BiggerGait: Unlocking Gait Recognition with Layer-wise Representations from Large Vision Models
Authors: Dingqiang Ye, Chao Fan, Zhanbo Huang, Chengwen Luo, Jianqiang Li, Shiqi Yu, Xiaoming Liu
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Comprehensive evaluations on CCPG, CAISA-B*, SUSTech1K, and CCGR_MINI validate the superiority of Bigger Gait across both withinand cross-domain tasks, establishing it as a simple yet practical baseline for gait representation learning. |
| Researcher Affiliation | Academia | 1 Department of Computer Science and Engineering, Southern University of Science and Technology 2 School of Artificial Intelligence, Shenzhen University 3 Department of Computer Science and Engineering, Michigan State University EMAIL, EMAIL, EMAIL, EMAIL, EMAIL, EMAIL |
| Pseudocode | No | The paper describes the Bigger Gait as "A Layer-wise LVM-based Gait Baseline" in Section 3.1 and illustrates its overview in Figure 2, but it does not contain a structured pseudocode or algorithm block. |
| Open Source Code | Yes | All the models and code are available at https://github.com/ShiqiYu/OpenGait/. |
| Open Datasets | Yes | Subsequent experiments are mainly conducted on four widely used clothing-variation and multi-view gait datasets: CCPG [33], CASIA-B* [75], SUSTech1K [56], and CCGR_MINI [80]. |
| Dataset Splits | Yes | Every experiment strictly follows the official protocols released by the owner. |
| Hardware Specification | Yes | Using eight 24GB RTX 6000 GPUs, the DINOv2-S-based Bigger Gait requires approximately 8.8 hours to train on CCPG. |
| Software Dependencies | No | The paper mentions the use of DINOv2, CLIP, and SAM, but does not provide specific version numbers for these models or any other software libraries/frameworks (e.g., PyTorch, TensorFlow, Python) used for implementation. |
| Experiment Setup | Yes | All input frames are resized to 448 224 for DINOv2 [48], 224 224 for CLIP [52] and 512 256 for SAM [30]. The training runs for 30k iterations using SGD (momentum = 0.9, weight-decay = 5 10 4) with an initial learning rate of 0.1, which is dropped by 10 at 15k and 25k steps. Each mini-batch adopts the tuple (p, k, l) = (8, 4, 30), which is 8 identities, 4 sequences per identity, and 30 frames per sequence. Frame sampling follows the protocol of Gait Base [13], and the sole augmentation is a random horizontal flip applied consistently to every frame within a sequence. |