Image Classification is a valid choice of algorithm, depending on the specifics on the use case under evaluation, but the Parquet file format is not a supported or appropriate choice here.
Object2Vec, although a multi-purpose algorithm, is primary used for to create embeddings for feature engineering.
Semantic segmentation, the ability to map instances of the same object, is the correct choice. Note this is different from instance segmentation (not currently supported) where specific instances of cars, chairs etc would be accessible.
Neural Topic Modelling, is an unsupervised learning algorithm that can be used to classify or summarize documents based on the topics detected or to retrieve information or recommend content based on topic similarities. Note AWS in built algorithms also include Latent Dirichlet Allocation (LDA), another unsupervised algorithm used for similar purposes.
No responses yet