DISCOVERING VISUAL CONCEPTS FROM WEAKLY LABELED IMAGE COLLECTIONS

摘要

Images uploaded to photo sharing websites often include some tags or sentence descriptions. In an example embodiment, these tags or descriptions, which might be relevant to the image contents, become the weak labels of these images. The weak labels can be used to identify concepts for the images using an iterative hard instance learning algorithm to discover visual concepts from the label and visual feature representations in the weakly labeled images. The visual concept detectors can be directly applied to concept recognition and detection.

主权项

1. A system comprising:
a memory having instructions embodied thereon; a processor configured by the instructions to perform operations comprising:
accessing a weakly-labeled dataset comprising a set of images, each image of the set of images having a set of associated weak labels;clustering images sharing an associated weak label;accessing a set of examples selected from the set of images, each example of the set of examples not being associated with the weak label associated with the clustered images;training a recognition engine using the clustered images as positive examples and the set of examples as negative examples; andgenerating strong labels for the weakly-labeled dataset, using the trained recognition engine.