Gradient Corner Pooling for Keypoint-Based Object Detection
Authors: Xuyang Li, Xuemei Xie, Mingxuan Yu, Jiakai Luo, Chengwei Rao, Guangming Shi
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We verify the gradient corner pooling algorithm on the dataset and in real scenarios, respectively. The networks with gradient corner pooling located the corner points earlier in the training process and achieve an average accuracy improvement of 0.2%-1.6% on the MS-COCO dataset. The detectors with gradient corner pooling show better angle adaptability for arrayed objects in the actual scene test. We test our method on the dataset and real-world scenarios respectively and conduct ablation experiments for each module. |
| Researcher Affiliation | Academia | Xuyang Li1, Xuemei Xie1, 2*, Mingxuan Yu1, Jiakai Luo1, Chengwei Rao1, Guangming Shi1, 3 1 Xidian University, Xi an, 710071, China. 2 Pazhou Lab, Huangpu, 510555, China. 3 Peng Cheng Laboratory, Shenzhen, 518055, China. {xylee, mxuanyu, jkluo, cwrao}@stu.xidian.edu.cn, xmxie@mail.xidian.edu.cn, gmshi@xidian.edu.cn |
| Pseudocode | Yes | Algorithm 1: Gradient corner pooling |
| Open Source Code | No | The paper does not provide an explicit statement or link to open-source code for the described methodology. |
| Open Datasets | Yes | We evaluate our method on the MS-COCO dataset (Lin et al. 2014). |
| Dataset Splits | Yes | We follow common practice (Lin et al. 2017) and use the COCO trainval35k split (union of 80k images from train and a random 35k subset of images from the 40k image val split) for training networks such as Corner Net, Corner Net-Saccade, Center Net and Center Net++. The ablation studies and visualization experiments are performed on the corresponding validation set. |
| Hardware Specification | Yes | For all the experiments, due to limited resources, we train Corner Net-104, Center Net-52, Center Net++, Rep Points v2, Corner Net-Saccade on four Ge Force RTX3090 GPUs |
| Software Dependencies | No | The paper states 'follow the training details in (Law and Deng 2018; Duan et al. 2019, 2022; Chen et al. 2020; Law et al. 2019), respec-tively.', but does not list specific software dependencies with version numbers. |
| Experiment Setup | No | The paper states 'follow the training details in (Law and Deng 2018; Duan et al. 2019, 2022; Chen et al. 2020; Law et al. 2019), respec-tively.', indicating that specific experimental setup details (like hyperparameters) are referenced from other papers rather than explicitly stated within this paper's main text. |