चलती_मनिस्ट

  • विवरण :

हस्तलिखित अंकों के MNIST डेटाबेस का मूविंग वेरिएंट। यह मॉडल प्रदर्शन की रिपोर्ट करने के लिए लेखकों द्वारा उपयोग किया जाने वाला डेटा है। MNIST डेटासेट से प्रशिक्षण/सत्यापन डेटा उत्पन्न करने के लिए tfds.video.moving_mnist.image_as_moving_sequence देखें।

विभाजित करना उदाहरण
'test' 10,000
  • फ़ीचर संरचना :
FeaturesDict({
    'image_sequence': Video(Image(shape=(64, 64, 1), dtype=uint8)),
})
  • फ़ीचर दस्तावेज़ीकरण :
विशेषता कक्षा आकार डीटाइप विवरण
विशेषताएं डिक्ट
image_sequence वीडियो (छवि) (20, 64, 64, 1) uint8
  • उद्धरण :
@article{DBLP:journals/corr/SrivastavaMS15,
  author    = {Nitish Srivastava and
               Elman Mansimov and
               Ruslan Salakhutdinov},
  title     = {Unsupervised Learning of Video Representations using LSTMs},
  journal   = {CoRR},
  volume    = {abs/1502.04681},
  year      = {2015},
  url       = {http://arxiv.org/abs/1502.04681},
  archivePrefix = {arXiv},
  eprint    = {1502.04681},
  timestamp = {Mon, 13 Aug 2018 16:47:05 +0200},
  biburl    = {https://dblp.org/rec/bib/journals/corr/SrivastavaMS15},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}