A diverse database of over 1.4B 3x3 convolution filters extracted from CNN models trained for various tasks in diverse image domains. We collected a total of 647 publicly available CNN models that have been pre-trained for various 2D visual tasks. In order to provide a heterogeneous and diverse representation of convolution filters "in the wild", we retrieved pre-trained models for 11 different tasks e.g. such as classification, segmentation} and image generation. We also recorded various meta-data such as depth and frequency of included operations for each model, and manually categorized the variety of used training sets into 16 visually distinctive groups like natural scenes, medical ct, seismic, or astronomy. In total, the models were trained on 71 different data sets. The dominant subset is formed by image classification models trained on ImageNet1k (355 models). More details: https://github.com/paulgavrikov/cnn-filter-db

This work was supported in part by the German Ministry for Science, Research and Arts Baden-Wuerttemberg (MWK) under Grant 32-7545.20/45/1 Quality Assurance of Machine Learning Applications (Q-AMeLiA). https://q-amelia.in.hs-furtwangen.de/