Brandon Castellano
Brandon Castellano
**Description of Problem & Solution** When scanning multiple videos, it would be useful if a log message could be emitted before each video is loaded. This would help debug issues...
Text and box overlays are supported by ffmpeg filters: https://ffmpegbyexample.com/examples/z1jezi81/drawtext_and_drawbox_using_sendcmd_manifest/ It should be possible to do the same but using a temporary file for the bounding box.
This PR changes how event merging works, fixing #72 and #192: - Event merging is no longer controlled by the `time-before-event` (`-tb`) and `time-post-event` (`-tp`) options - These continue to...
Investigate a few different modes of parallelization of scans (in order of general difficulty): 1. scanning multiple videos that are independent 2. scanning multiple videos being concatenated with multiple readers,...
Right now min scene length is set to 2 frames, which may be undesirable at higher framerates. Most test footage is between 10 and 60 frames per second, so say...
Right now kernel size is corrected for the downscale factor of the image even if set explicitly. This should probably not be done if kernel size is set explicitly.
There should be a set of tests for DVR-Scan that test the behaviour on entire videos. These tests can be a separate job that runs perhaps on a weekly basis...
In v1.6 there seems to be a bug with the progress bar length when setting the start time. This doesn't affect processing accuracy or output. However, it does mean the...
As per the discussion in #184 there are some opportunities to improve parameter logging. For example right now when you run DVR-Scan, we report the parameters of the background subtractor,...
It looks like in v0.6.x after `export-html` was renamed to `save-html`, but backwards compatibility was kept with the prior. We emit the warning to use the new command name with...