Awesomes
Images
- Pascal VOC
- CIFAR-10 10个类别ML基准测试集
- ImageNet
- SVHN
- WMT
- openimages
- FlickrStyle Flickr Style
- RN1015k500 picture's location
- VisualSentimentCNN
- MobileNet CoreML Demo
- GoogLeNetPlaces image scene(205 categories such as airport, bedroom, forest, coast etc.)
- Inceptionv3 the dominant objects in an image (1000 categories such as trees, animals, food, vehicles, person etc.)
- Resnet50
- VGG16
- TinyYOLO
Person(Face/Age/Emotions/Gender..)
- 香港的CelebA
- WIDERFACE
- AFW
- SUN SUN 负样本
- AgeNet Age
- GenderNet Gender
- CNNEmotions Emotions
- Nudity NSFW (nude) or SFW (not nude)
Car
- CarRecognition Predict the brand & model of a car.
- Vehicle Color Dataset car color
- BIT-Vehicle Dataset car type
Flower
Food
Handwritten
- MNIST handwritten digit
NLP/Text Analysis
- SentimentPolarity
- DocumentClassification
- MessageClassifier
- NamesDT
- Maluuba Datasets
- Billion Words
- Common Crawl
Geospatial
- OpenStreetMap OpenStreetMap data
- Landsat8 The Landsat 8 satellite images(卫星拍摄的地球表面数据)
- NEXRAD The Next Generation Weather Radar（国大气环境数据）
others
Voice
- YouTube-8M 使用 4,716 个不同类别标记超过 700 万个 YouTube 视频
- YouTube-Bounding Boxes 500 万个边界框，涉及 210,000 个 YouTube 视频
- Speech Commands Dataset 简短命令语，涉及数千个人
- AudioSet 使用 527 个不同的声音事件标记 200 万个时长为 10 秒的 YouTube 剪辑
- 原子视觉动作 AVA 21 万个动作标签，涉及 57,000 个视频剪辑
- Open Images 使用 6000 个类别标记 900 万个创意共享许可图像
- 带边界框的 Open Images 120 万个边界框，适合 600 个类别
