MetaJND: A Meta-Learning Approach for Just Noticeable Difference Estimation
Authors: Miaohui Wang, Yukuan Zhu, Rong Zhang, Wuyuan Xie
IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results on four benchmark datasets demonstrate the effectiveness of our Meta JND. Moreover, we have also evaluated its performance in compression and watermarking applications, observing higher bit-rate savings and better watermark hiding capabilities. |
| Researcher Affiliation | Academia | 1College of Computer Science and Software Engineering, Shenzhen University 2Guangdong Key Laboratory of Intelligent Information Processing, Shenzhen University |
| Pseudocode | No | No explicit pseudocode or algorithm blocks were found in the paper. The methods are described textually and through diagrams. |
| Open Source Code | No | The paper does not provide a statement or link indicating that the source code for the described methodology is publicly available. |
| Open Datasets | Yes | In the experiments, we have trained the Meta JND using the benchmark dataset PWJND [Shen et al., 2020]. To verify the generalization ability of different models, we further select three additional JND benchmark datasets, including MCL-JCI [Wang et al., 2016], Kon JND-1k [Lin et al., 2022], and MDTJND [Liu et al., 2023] for testing. |
| Dataset Splits | Yes | We randomly split the whole dataset into training, validation, and test sets with a ratio of 8:1:1. |
| Hardware Specification | Yes | The model training is then conducted on NVIDIA Ge Force RTX 3090 GPU, which takes approximately 16 hours for 200 epochs. |
| Software Dependencies | No | Our Meta JND-Net is implemented on the Py Torch with all weights initialized using a truncated normal initializer. We use the default parameters of the Adam optimizer, such as β1 = 0.9 and β2 = 0.999. |
| Experiment Setup | Yes | Before the training, we crop the input images into 224 224 with a random cropping and a random rotation. We set the batch size to 8 and the initial learning rate to 1e-5. To guarantee the alignment efficiency, we have developed a meta-alignment loss Lalign, which is calculated by a pixelwise L1 distance between the input RGB feature Frgb and the aligned feature ˆFy. With α setting to 1, the total loss function in (1) can be expressed as: Loverall = X y dep,sal Frgb ˆFy 1 + α Igt Ijnd 2 2 |