Detecting Human-Object Interactions with Object-Guided Cross-Modal Calibrated Semantics

Authors: Hangjie Yuan, Mang Wang, Dong Ni, Liangpeng Xu3206-3214

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments conducted on two popular HOI detection benchmarks demonstrate the significance of incorporating the statistical prior knowledge and produce state-of-the-art performances.
Researcher Affiliation Collaboration Hangjie Yuan1, Mang Wang3, Dong Ni1,2*, Liangpeng Xu3 1 College of Control Science and Engineering, Zhejiang University, Hangzhou, China 2 State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou, China 3 DAMO Academy, Alibaba Group, China
Pseudocode No The paper does not contain structured pseudocode or algorithm blocks.
Open Source Code Yes The codes are available at https://github.com/Jacob Yuan7/OCNHOI-Benchmark.
Open Datasets Yes In this paper, we use two widely-adopted datasets dubbed HICO-DET (Chao et al. 2015) and V-COCO (Gupta and Malik 2015).
Dataset Splits Yes HICO-DET contains 37,536 training images and 9,515 testing images... V-COCO contains 2,533 training images, 2,867 validating images and 4,946 testing images...
Hardware Specification Yes Computational cost analysis on HICO-DET with Tesla V100.
Software Dependencies No The paper does not provide specific software dependencies with version numbers, such as Python or library versions (e.g., PyTorch 1.9, CUDA 11.1).
Experiment Setup Yes We adopt Adam W (Loshchilov and Hutter 2018) to optimize OCN for 80 epochs with a weight decay of 10^-4. The learning rate (lr) of the backbone is fixed to 10^-5. The lr of other parts starts from 10^-4 and decays to 10^-5 after the 60th epoch. ... We set number of queries Nq = 100 and the dimension of queries D = 256. The head number H for Inter C and Intra EC is set to 2. The temperature τ of LSKL is set to 0.05. The smoothing parameter β is set to 0.1. ... The hyper-parameters λ1, λ2, λ3, λ4, λ5 in L are set to 1, 2.5, 1, 1, 1 respectively.