Just Noticeable Visual Redundancy Forecasting: A Deep Multimodal-Driven Approach
Authors: Wuyuan Xie, Shukang Wang, Sukun Tian, Lirong Huang, Ye Liu, Miaohui Wang
AAAI 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results on eight different benchmark datasets validate the superiority of our hm JND-Net over eight representative methods. |
| Researcher Affiliation | Academia | 1 Guangdong Key Laboratory of Intelligent Information Processing, Shenzhen University 2 Peking University 3 Nanjing University of Posts and Telecommunications |
| Pseudocode | No | The text does not contain structured pseudocode or algorithm blocks. |
| Open Source Code | No | The text does not provide concrete access to source code for the methodology described in this paper. |
| Open Datasets | Yes | We trained hm JND-Net based on the latest benchmark dataset (Shen et al. 2020), which covers various image contents, including outdoor, indoor, landscape, nature, people, objects and buildings. The dataset consists of 202 high-definition original images with the size of 1920 1080 and 7878 redundancy-removed images by VVC. |
| Dataset Splits | Yes | We randomly select image pairs from the dataset as the training set, validating set, and testing set based on the ratio of 8:1:1. |
| Hardware Specification | Yes | We have trained the hm JND-Net with a mini-batch size of 16 for 200 epochs on an Nvidia Tesla A100 GPU, which takes about 10 hours. |
| Software Dependencies | No | The paper mentions 'Py Torch' but does not provide a specific version number, nor does it list other software dependencies with their versions. |
| Experiment Setup | Yes | We have trained the hm JND-Net with a mini-batch size of 16 for 200 epochs on an Nvidia Tesla A100 GPU, which takes about 10 hours. The initial learning rate is 1e-4, which will linearly decay to 0 from the 200-th epoch. Due to the large image size (1920 1080) in the dataset, an adaptive partition is performed on all input images, with the partition size of 224 224. |