Neural data sets


 * http://commoncrawl.org/the-data/ entire internet archived, available for purchase.
 * https://www.reddit.com/r/dldata/, https://www.reddit.com/r/MachineLearning/ , https://www.reddit.com/r/datamining/
 * https://www.reddit.com/r/dldata/ links to https://graphics.ethz.ch/~perazzif/davis/code.html, https://github.com/fperazzi/davis DAVIS (Densely Annotated VIdeo Segmentation), consists of fifty high quality, Full HD video sequences, spanning multiple occurrences of common video object segmentation challenges such as occlusions, motion-blur and appearance changes. Each video is accompanied by densely annotated, pixel-accurate and per-frame ground truth segmentation. https://graphics.ethz.ch/%7Eperazzif/davis/index.html for youtube video links. cv-foundation.org Perazzi academic paper.
 * https://www.tensorflow.org/
 * https://research.googleblog.com/2016/06/wide-deep-learning-better-together-with.html  https://research.googleblog.com/2016/09/introducing-open-images-dataset.html, https://github.com/openimages/dataset Today, we introduce Open Images, a dataset consisting of ~9 million URLs to images that have been annotated with labels spanning over 6000 categories. We tried to make the dataset as practical as possible: the labels cover more real-life entities than the 1000 ImageNet classes, there are enough images to train a deep neural network from scratch. https://research.googleblog.com/2015/07/deepdream-code-example-for-visualizing.html
 * http://megaface.cs.washington.edu/dataset/download.html 65gig data set of million faces, bounding boxes. See reddit/r/dldata
 * http://places.csail.mit.edu/demo.html This demo identifies if the image is an indoor or an outdoor place, and suggests the five most likely place categories representing the image, using Places-CNN (see project page). It is made for pictures of environments, places, views on a scene and a space (as opposed to picture of an object). You also could upload image using mobile phone. Upload .jpg or jpeg image only. The heatmap is generated by the CAM technique.
 * http://mi.eng.cam.ac.uk/projects/relocalisation/#dataset, https://github.com/alexgkendall/caffe-posenet POSENET Alex Kendall, Matthew Grimes and Roberto Cipolla "PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization." Proceedings of the International Conference on Computer Vision (ICCV), 2015.
 * The main goal of this challenge is to recognize objects from a number of visual object classes in realistic scenes (i.e. not pre-segmented objects). It is fundamentally a supervised learning learning problem in that a training set of labelled images is provided. The twenty object classes that have been selected are: http://host.robots.ox.ac.uk/pascal/VOC/voc2012/index.html#devkit

links
Opencv