Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].
Texture Networks: Feed-forward Synthesis of Textures and Stylized Images
Authors: Dmitry Ulyanov, Vadim Lebedev, Andrea, Victor Lempitsky
ICML 2016 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We compare our method to (Gatys et al., 2015a;b) using the popular implementation of (Johnson, 2015), which produces comparable if not better results than the implementation eventually released by the authors. We also compare to the DCGAN (Radford et al., 2015) version of adversarial networks (Goodfellow et al., 2014). [...] Figure 4 shows the results obtained by the four methods on two challenging textures of (Portilla & Simoncelli, 2000). Qualitatively, our generator CNN and (Gatys et al., 2015a) s results are comparable and superior to the other methods; however, the generator CNN is much more efficient (see Sect. 4.2). |
| Researcher Affiliation | Collaboration | Dmitry Ulyanov EMAIL Computer Vision Group, Skoltech & Yandex, Russia Vadim Lebedev EMAIL Computer Vision Group, Skoltech & Yandex, Russia Andrea Vedaldi EMAIL Visual Geometry Group, University of Oxford, United Kingdom Victor Lempitsky EMAIL Computer Vision Group, Skoltech, Russia |
| Pseudocode | No | No structured pseudocode or algorithm block was found in the paper. |
| Open Source Code | No | The paper refers to an 'Extended version' at http://arxiv.org/abs/1603.03417 and 'Supp.Material' but does not contain an explicit statement from the authors providing concrete access to their own source code for the described methodology. It mentions using 'Johnson, 2015. neural-style. https://github.com/jcjohnson/neural-style' which is a third-party implementation, not their own code release. |
| Open Datasets | Yes | Our loss function is derived from (Gatys et al., 2015a;b) and compares image statistics extracted from a fixed pretrained descriptor CNN (usually one of the VGG CNN (Simonyan & Zisserman, 2014; Chatfield et al., 2014) which are pre-trained for image classification on the Image Net ILSVRC 2012 data). [...] For training, example natural images were extracted at random from the Image Net ILSVRC 2012 data. |
| Dataset Splits | No | The paper mentions using 'Image Net ILSVRC 2012 data' for training but does not specify exact train/validation/test dataset splits, percentages, or sample counts for reproducibility. |
| Hardware Specification | No | The paper mentions that the iterative procedure requires 'several seconds in order to generate a relatively small image using a high-end GPU', but it does not specify the exact model or type of GPU, CPU, or other hardware used for their experiments. |
| Software Dependencies | No | The paper mentions 'Torch7 s implementation of Adam (Kingma & Ba, 2014)' but does not provide specific version numbers for Torch7 or other key software dependencies (e.g., libraries, frameworks) required to replicate the experiments. |
| Experiment Setup | Yes | The generator network weights were initialized using Xavier s method. Training used Torch7 s implementation of Adam (Kingma & Ba, 2014), running it for 2000 iteration. The initial learning rate of 0.1 was reduced by a factor 0.7 at iteration 1000 and then again every 200 iterations. The batch size was set to 16. Similar to (Gatys et al., 2015a), the texture loss uses the layers LT = {relu1 1, relu2 1, relu3 1, relu4 1, relu5 1} of VGG-19 and the content loss the layer LC = {relu4 2}. |