Computer Vision applications rely of large volumes of video data to be processed to learn the patterns related to the objects/regions of interest. From robotic vision to object detection and real-time object tracking applications for autonomous drive, there is a need to isolate quality image frames from long sequence of videos that can then train respective machine learning (ML) applications. The video-data processing pipeline can be combined with modeling and deployment pipelines specifically for video/image-based ML applications.
The yaml file contains the following metadata of the video:
- Number of original frames in the Video File when first split
- The frames removed during the Laplacian filter
- Laplacian variance spread of frames sent through the Laplacian filter
- The frames removed during the Structural Similarity filter
- Structural similarity spread of frames sent through the Structural Similarity filter
- Ratio and absolute number of frames removed during the Laplacian filter
- Ratio and absolute number of frames removed during the Structural Similarity filter
- Number of detected objects and classification counts of detected objects on the filtered frames by frame
- Scene classification of each frame
Examples of the Output File can be found HERE
Example of yaml output
Please get in touch with me at [email protected] if you have any questions or inquiries.