Attribute Aware Pooling for Pedestrian Attribute Recognition
Authors: Kai Han, Yunhe Wang, Han Shu, Chuanjian Liu, Chunjing Xu, Chang Xu
IJCAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on benchmark datasets demonstrate that the proposed pooling method appropriately explores and exploits the correlations between attributes for the pedestrian attribute recognition. |
| Researcher Affiliation | Collaboration | Kai Han1, Yunhe Wang1, Han Shu1, Chuanjian Liu1, Chunjing Xu1, Chang Xu2 1Huawei Noah s Ark Lab 2School of Computer Science, FEIT, University of Sydney, Australia |
| Pseudocode | Yes | Algorithm 1 Attribute aware pooling method for pedestrain attribute classification with CNNs. |
| Open Source Code | No | The paper does not provide any explicit statement or link for open-source code for the described methodology. |
| Open Datasets | Yes | The evaluation is conducted on three largest publicly available pedestrian attribute datasets: PETA [Deng et al., 2014], RAP [Li et al., 2016], and PA-100K [Liu et al., 2017b] |
| Dataset Splits | Yes | PETA [Deng et al., 2014]: ...we randomly divide the whole dataset into three partitions: 9,500 images for training, 1,900 for validation, and 7,600 for testing... PA-100K [Liu et al., 2017b]: ...The whole dataset is randomly split into training, validation and test sets with a ratio of 8 : 1 : 1. |
| Hardware Specification | Yes | All methods were implemented using Py Torch [Paszke et al., 2017] and run on NVIDIA V100 graphics cards. |
| Software Dependencies | No | All methods were implemented using Py Torch [Paszke et al., 2017] and run on NVIDIA V100 graphics cards. While PyTorch is mentioned, a specific version number is not provided, which is required for reproducibility. |
| Experiment Setup | Yes | Adam optimizer [Kingma and Ba, 2014] with a batch size of 32 is used to fine-tune the entire model with an initial learning rate of 0.01. The images are resized to 512 × 256 and during training randomly cropped to the size of 448 × 224 with random horizontal flipping. We tested the impact of λ by tuning it from 0 to 0.5 with the step of 0.05 on the m A metric using the PETA dataset. |