Autoencoder Image Interpolation by Shaping the Latent Space
Authors: Alon Oring, Zohar Yakhini, Yacov Hel-Or
ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We tested our method against two different datasets: the synthetic pole dataset, which was rendered using the Unity game engine, where all images were taken from a fixed viewing position (from above) and the COIL-100 dataset. For the first dataset, a single illumination source was rotated at intervals of 5 degrees along the azimuth at different altitudes, ranging from 45 to 80 degrees with respect to the plane in 5-degree intervals. This dataset contains a total of 576 images. In the second dataset, to test our method against real images with complex geometric and photometric parameterization, we used the COIL-100 dataset (Nene et al., 1996) containing color images of 100 objects. |
| Researcher Affiliation | Academia | 1School of Computer Science, The Interdisciplinary Center, Herzliya, Israel . Correspondence to: Y.H. <toky@idc.ac.il>. |
| Pseudocode | No | The paper describes the architecture and implementation steps, but it does not include formal pseudocode or an algorithm block. |
| Open Source Code | No | The paper does not contain any statement about releasing source code, nor does it provide a link to a code repository. |
| Open Datasets | Yes | We tested our method against two different datasets: the synthetic pole dataset, which was rendered using the Unity game engine, where all images were taken from a fixed viewing position (from above) and the COIL-100 dataset. [...] we used the COIL-100 dataset (Nene et al., 1996) containing color images of 100 objects. |
| Dataset Splits | No | A random 80%-20% training-testing split was chosen for all experiments, using the same batch size and total number of examples in the dataset. While 'validation dataset' is mentioned later (Section 4.2), its specific split percentage or sample count is not provided in the general splitting description. |
| Hardware Specification | Yes | All experiments were performed using a single NVIDIA V100 GPU. |
| Software Dependencies | No | The paper mentions using a 'VGG-inspired' architecture but does not specify version numbers for any software dependencies such as libraries or programming languages used (e.g., Python, PyTorch, TensorFlow versions). |
| Experiment Setup | Yes | During log grid-search hyperparameter optimization, we found that λA = λC = 10^-2 and λS = 10^-1 produce the best results. |