Skip to content

Latest commit

 

History

History
13 lines (13 loc) · 828 Bytes

prod_frame_train_vid.md

File metadata and controls

13 lines (13 loc) · 828 Bytes

Effect

Overesteemed results by a model that restores video and make prediction as a whole. However, its performance significantly drops when it operates in an online regime, predicting one frame at a time.

Rules to avoid leakage

Represent data for training in same modality as it will be inferenced in production, if each frame need to be predicted stratify train dataset to break the ability to interpolate prediction from frames inaccesible in production mode of work.

Incorporation stage

Ground truth gathering. Dataset preparation.

Locate in code

Сroping on frames, selection which frames to include in train dataset

Was met or loosely based on

kaggle "State Farm Distracted Driver Detection" competition JACOBKIE solution