aurelio-ai/xclip-base-patch16-zero-shot

reverted back

updated hugginface hub

updating transformer version

update hugginface hub

Added a variable to handle OCR extraction frame rate

Instead of OCR every frame, do it every 6 video frames

Instead of OCR every frame, doit every 8 video frames

Changed back to set, with jsondumps for serializing

Changed set for list to be json serializable

Added easyocr for videoframes

Removed some loggings

Added back some logging messages

Added back some logging messages

Limited batch size to 4

Returned to ThreadPoolExecutor

Changed ThreadPoolExecutor for ProcessPoolExecutor

Changed OpenCV2 for decord in extracting videoframes

Changed OpenCV2 for decord in extracting videoframes

Added timeit loggings

Added timeit loggings

Modified the squeeze() application to embeddings

Added normalization using torch and shape check

Added normalization using torch

Removed normalization of embeddings

Removed troch.stack()

Modified embed_frames_with_xclip_processing

Added more logging messages

Added logging messages

Initial commit of XCLIP video processing endpoint only video frames

Duplicate from microsoft/xclip-base-patch16-zero-shot