Video labeling method and system based on image clustering and electronic equipment

Abstract

The invention relates to the technical field of data acquisition, in particular to a video labeling method and system based on image clustering and electronic equipment, and the method comprises the steps: S1, a video is provided and the detection model reads the first frame of image in the video and confirms the plurality of targets needing to be tracked; The target tracking model reads each frame of image in the video and identifies a plurality of targets needing to be tracked in each frame of image to obtain a plurality of images; feature extraction is carried on on each target in each image by using a feature extractor to obtain clustering feature information of each target; Clustering targets of the same category in the plurality of images by using the clustering feature information through a clustering algorithm; Compared with the prior art, the method has the advantages that multiple frames of images are acquired, multiple targets are labeled, multiple targets of the same category in the multiple frames of images are clustered through a clustering algorithm, only one of the targets of the same category needs to be labeled when the targets are labeled, each target does not need to be labeled separately, and the labeling efficiency can be well improved.

Description

【Technical field】 [0001] The present invention relates to the technical field of data collection, in particular to a video tagging method, system and electronic equipment based on image clustering. 【Background technique】 [0002] Nowadays, industries such as goods transportation, goods production, and goods sales all need to classify and detect different types of goods based on the characteristic information of the goods themselves. Therefore, when faced with a large number of goods that need to be detected and classified, it is usually necessary Annotate multiple feature data information of a cargo in multiple directions and in different scenes. When annotating a video, it is often necessary to annotate the position and classification of objects in each frame. There are many redundant tasks, which are time-consuming and labor-intensive. long. 【Content of invention】 [0003] In order to overcome the technical problem of low labeling efficiency in the current data collecti...

AI Technical Summary

Problems solved by technology

[0003] In order to overcome the technical problem of low labeling efficiency in the current data collection process, the present invention provides a video labeling method, system and electronic equipment based on image clustering

Images