[FFmpeg-devel] [PATCH V7 4/6] lavu: add side data AV_FRAME_DATA_BOUNDING_BOXES

Guo, Yejun yejun.guo at intel.com
Mon Apr 12 05:18:56 EEST 2021



> -----Original Message-----
> From: ffmpeg-devel <ffmpeg-devel-bounces at ffmpeg.org> On Behalf Of
> Pedro Arthur
> Sent: 2021年4月12日 5:12
> To: FFmpeg development discussions and patches
> <ffmpeg-devel at ffmpeg.org>
> Subject: Re: [FFmpeg-devel] [PATCH V7 4/6] lavu: add side data
> AV_FRAME_DATA_BOUNDING_BOXES
> 
> Em dom., 11 de abr. de 2021 às 14:53, Nicolas George <george at nsup.org>
> escreveu:
> >
> > Anton Khirnov (12021-04-11):
> > > We are a generic multimedia framework. "the field" for us is
> multimedia
> > > in general, so we should use names meaningful in general multimedia
> > > context.
> > > I mostly agree with Lynne, "bounding box" is confusing and misleading
> > > when this structure is built around object detection and classification.
> >
> > I agree with both of you. When faced with this kind of choice, we must
> > choose the wording that will be as clear and not confusing as possible
> > for people familiar with general concepts of video encoding but not
> > necessarily familiar with the jargon of any particular sub-field. The
> > specialists of the particular subfield are supposed to be more capable
> > of adjusting.

Thanks both, got the point.

> >
> Personally, "bounding box" is very clear to me, I might be biased.
> It seems we are bikeshedding over a naming. I think it is more
> constructive if we propose a better name as the only other option
> proposed until now seems worst to me.
> 
> I think something like "AV_SIDE_DATA_DETECTION_BOUNDING_BOX" is
> reasonable, as it conveys what it is and what it is used for.
> 

Agree, thanks. For people who use this feature will feel a bit
strange if they do not see bounding box or bbox. (bbox is the
abbreviation of bounding box).

We might use BBOX to replace BOUNDING_BOX since it is too long,
and people in the sub-filed know BBOX as well. I'm open to change
it back. And I also add 'ES' as the suffix since it is an array. So, the
code might looks like:

frame.h:
    /**
     * Bounding boxes for object detection and classification, ...
     */
    AV_FRAME_DATA_DETECTION_BBOXES,

detection_bbox.h/c:
struct AVDetectionBbox;
struct AVDetectionBboxHeader;

av_get_detection_bbox
av_detection_bboxes_alloc
av_detection_bboxes_create_side_data



More information about the ffmpeg-devel mailing list