A Probabilistic U-Net for Segmentation of Ambiguous Images
Authors: Simon Kohl, Bernardino Romera-Paredes, Clemens Meyer, Jeffrey De Fauw, Joseph R. Ledsam, Klaus Maier-Hein, S. M. Ali Eslami, Danilo Jimenez Rezende, Olaf Ronneberger
NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We show on a lung abnormalities segmentation task and on a Cityscapes segmentation task that our model reproduces the possible segmentation variants as well as the frequencies with which they occur, doing so significantly better than published approaches. |
| Researcher Affiliation | Collaboration | 1Deep Mind, London, UK 2Division of Medical Image Computing, German Cancer Research Center, Heidelberg, Germany {simon.kohl,k.maier-hein}@dkfz.de {brp,meyerc,defauw,jledsam,aeslami,danilor,olafr}@google.com |
| Pseudocode | No | The paper does not contain any explicitly labeled "Pseudocode" or "Algorithm" blocks or figures. |
| Open Source Code | Yes | An open source re-implementation of our approach can be found at https://github.com/Simon Kohl/probabilistic_unet. |
| Open Datasets | Yes | Here we consider two datasets: The LIDC-IDRI dataset [32, 33, 34] which contains 4 annotations per input, and the Cityscapes dataset [35] |
| Dataset Splits | Yes | For our experiments we split this dataset into a training set composed of 722 patients, a validation set composed of 144 patients, and a test set composed of the remaining 144 patients. ... and split off 274 images (corresponding to the 3 cities of Darmstadt, Mönchengladbach and Ulm) from the official training set as our internal validation set. |
| Hardware Specification | Yes | For our experiments we used 8 Tesla P100 GPUs. |
| Software Dependencies | No | The paper mentions using the "Adam optimizer [37]" but does not specify versions for programming languages (e.g., Python), deep learning frameworks (e.g., TensorFlow, PyTorch), or other libraries. |
| Experiment Setup | Yes | We train for 1500 epochs... using the Adam optimizer [37] with default parameters and a learning rate of 1e-4. We decay the learning rate by a factor of 10 at epochs 1000 and 1250. A batch size of 20 is used... |