Source: https://pan. Quark. Cn/s/1b76e6a3b570** title and description reading** "maximity of access to mscoco datasets" clearly points to our focus on the mscoco dataset and is linked to access. The description of “mscoco datasets, both in the 2014-2015 version” further clarifies that what we need is the mscoco datasets for 2014 and 2015. This data set plays a key role in the computer visual field and is widely used for a variety of task types, such as image classification, target detection, semantic division and image processing. ** mscoco data sets detailing **mscoco (msco common objects in text) is a large-scale multi-modular data set initiated by microsoft to facilitate research progress in computer visual and natural language processing. The data set contains a large number of images and their corresponding explanatory information, which cover a wealth of objects, scenes and behavioural descriptions and provide sufficient learning material for training in in-depth learning models. ** structure of data sets**1. ** image data* *: mscoco data sets contain more than 330,000 images, which are systematically divided into three parts: training, validation and testing. Each image was carefully marked, with an average of about five objects per image. 2. ** marked information**: each object instance is given precise boundary frame positioning and is classified into 80 predefined categories. In addition, a partition mask is provided, which is essential for carrying out semantic splitting tasks. 3. ** image description**: in addition to object labelling, each image is equipped with a description (caption) produced in five natural languages, a feature that gives the mscoco dataset a unique advantage in the image production task. 4. ** key points note**: for specific types of objects, such as humans, precise location information containing 28 key points is also provided for human attitude estimates and movements...
