v.1.41.0

Improvements:

1. [LUNA-8877] Human tracking stream analytics: removed the targets that are not needed for the events (the events are built by the handler from the face and body samples): overview, warp_quality, emotions, mask, glasses, eyes, gaze, mouth, aggregated_liveness, aggregated_deepfake, aggregated_face_thumbnail, aggregated_face_basic_attributes, aggregated_face_descriptor, aggregated_body_thumbnail, aggregated_body_attributes, aggregated_body_descriptor. The remaining targets are face_detection, landmarks5, head_pose, ags, aggregated_face_samples, body_detection, aggregated_body_samples. The overview and aggregated_* results of the removed targets were not delivered anyway, but the frame estimations warp_quality, emotions, mask, glasses, eyes, gaze, mouth were sent to the web socket subscribers in the estimated.face of each frame and are not sent anymore. Existing streams stay valid and track the same objects: a removed face target is replaced with face_detection, a removed body target with body_detection, overview with face_detection unless a body target is given. The image_retain_policy parameter is no longer used and is ignored. The human tracking analytics of videosdk is not affected.

Migration:

See a migration manual.