Learning Object Context for Dense Captioning
Authors: Xiangyang Li, Shuqiang Jiang, Jungong Han8650-8657
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experiments on benchmark datasets demonstrate the superiority of our proposed approach over the state-of-the-art methods. |
| Researcher Affiliation | Academia | 1Key Laboratory of Intelligent Information Processing of Chinese Academy of Sciences (CAS), Institute of Computing Technology, CAS, Beijing, 100190, China 2University of Chinese Academy of Sciences, Beijing, 100049, China 3School of Computing and Communications, Lancaster University, United Kingdom |
| Pseudocode | No | The paper does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide concrete access to source code for the methodology described in this paper. |
| Open Datasets | Yes | We use the Visual Genome (VG) dataset (Krishna et al. 2017) and the VG-COCO dataset which is the intersection of VG V1.2 and MS COCO (Lin et al. 2014) as the evaluation benchmarks. |
| Dataset Splits | Yes | The training, validation and test splits are the same with (Johnson, Karpathy, and Fei Fei 2016). There are 77,398 images for training and 5,000 images for validation and test respectively. There are 38,080 images for training, 2,489 images for validation and 2,476 for test. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments. |
| Software Dependencies | No | The paper mentions using VGG-16 and ResNet-101 as base networks and Faster R-CNN, but does not provide specific ancillary software details with version numbers (e.g., library or solver names). |
| Experiment Setup | Yes | The size of our vocabulary is 1000. The max length of all the sentence is set to 10. For all of the LSTM networks, the size of the hidden state is set to 512. We train the full model end-to-end in a single step of optimization. For the RPN, 12 anchors are used to generate possible proposals at each location and 256 boxes are sampled for each forward procedure. |