NeRV: Neural Representations for Videos

Authors: Hao Chen, Bo He, Hanyu Wang, Yixuan Ren, Ser Nam Lim, Abhinav Shrivastava

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform experiments on Big Buck Bunny sequence from scikit-video to compare our Ne RV with pixel-wise implicit representations, which has 132 frames of 720 1080 resolution. To compare with state-of-the-arts methods on video compression task, we do experiments on the widely used UVG [7], consisting of 7 videos and 3900 frames with 1920 1080 in total.
Researcher Affiliation Collaboration 1University of Maryland, College Park, 2Facebook AI
Pseudocode No The paper does not contain any pseudocode or clearly labeled algorithm blocks. Figure 3 illustrates a pipeline, but it is a diagram, not structured pseudocode.
Open Source Code Yes The source code and pre-trained model can be found at https://github.com/haochen-rye/Ne RV.git.
Open Datasets Yes We perform experiments on Big Buck Bunny sequence from scikit-video to compare our Ne RV with pixel-wise implicit representations, which has 132 frames of 720 1080 resolution. To compare with state-of-the-arts methods on video compression task, we do experiments on the widely used UVG [7], consisting of 7 videos and 3900 frames with 1920 1080 in total.
Dataset Splits No No explicit information on training/validation/test dataset splits, such as percentages or sample counts, was provided in the paper. The paper mentions 'training epochs' and 'batchsize' but does not define a separate validation set split.
Hardware Specification Yes All experiments are run with NVIDIA RTX2080ti.
Software Dependencies No The paper mentions 'Py Torch [54]' but does not provide a specific version number. Other mentioned tools like 'Adam optimizer [51]' and 'cosine annealing learning rate schedule [52]' are algorithms or schedules, not software dependencies with version numbers.
Experiment Setup Yes In our experiments, we train the network using Adam optimizer [51] with learning rate of 5e-4. For ablation study on UVG, we use cosine annealing learning rate schedule [52], batchsize of 1, training epochs of 150, and warmup epochs of 30 unless otherwise denoted. When compare with state-of-the-arts, we run the model for 1500 epochs, with batchsize of 6. For experiments on Big Buck Bunny , we train Ne RV for 1200 epochs unless otherwise denoted. For fine-tune process after pruning, we use 50 epochs for both UVG and Big Buck Bunny .