Smart labeling is where the pre-trained or base model helps to annotate the dataset prior to human annotation. The base model is the one we create with a small labeled dataset by using transfer learning.
There are multiple options, Export in TensorFlow / Pytorch’s standard saved model format Export as tf-serving docker image with http/grpc endpoints to serve.
Yes, SmartEye supports Image Classification, Object Detection, and Image Segmentation
Yes, In fact, SmartEye supports exporting annotations in all standard formats
SmartEye supports three types of data integration Direct data upload from local Cloud/Edge storage path mount for large datasets. Using public URL of video/Images 2nd option is helpful with large media files.