Progressive One-shot Human Parsing
Authors: Haoyu He, Jing Zhang, Bhavani Thuraisingham, Dacheng Tao1522-1530
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on the ATR-OS benchmark tailored for OSHP demonstrate POPNet outperforms other representative one-shot segmentation models by large margins and establishes a strong baseline. |
| Researcher Affiliation | Academia | Haoyu He, 1 Jing Zhang, 1 Bhavani Thuraisingham, 2 Dacheng Tao 1 1 The University of Sydney, Australia, 2 The University of Texas at Dallas, USA |
| Pseudocode | No | The paper does not contain any structured pseudocode or algorithm blocks. |
| Open Source Code | Yes | Source code can be found at https://github.com/Charleshhy/One-shot-Human-Parsing. |
| Open Datasets | Yes | In this session, we illustrate how to tailor the existing large-scale ATR dataset (Liang et al. 2015a,b) into a new ATR-OS dataset for the OSHP setting. |
| Dataset Splits | No | The paper describes how the ATR-OS dataset is split into training (Qtrain, Strain) and testing (Qtest, Stest) sets for the meta-learning setup. However, it does not explicitly mention or define a separate 'validation' dataset split for hyperparameter tuning or model selection in the conventional sense. |
| Hardware Specification | Yes | In this paper, we conduct the experiments on a single NVIDIA Tesla V100 GPU. |
| Software Dependencies | No | The paper mentions using an SGD optimizer and a backbone network pretrained on the COCO dataset, but it does not specify software dependencies with version numbers (e.g., PyTorch version, Python version, CUDA version). |
| Experiment Setup | Yes | The images are resized to 576 576 in one-way OSHP tasks and resized to 512 512 in k-way tasks... Training images are augmented by a random scale from 0.5 to 2, random crop, and random flip. We train the model using the SGD optimizer for 30 epochs with the poly learning rate policy. The initial learning rate is set to 0.001 with batch size 2. When generating dynamic prototypes, α in Eq. (2) is set to 0.001 by grid search. |