Skip to content

Update main branch w/ demo version#49

Merged
cwlacewe merged 21 commits into
mainfrom
cwl_demo
Jun 11, 2026
Merged

Update main branch w/ demo version#49
cwlacewe merged 21 commits into
mainfrom
cwl_demo

Conversation

@cwlacewe

@cwlacewe cwlacewe commented Jun 9, 2026

Copy link
Copy Markdown
Contributor

The overall flow of application has changed and now leverages fastapi.
This version is ready for demo and can be merged into main.

There is still room for improvement:

  • Investigate any instances (if any) of thread deadlocks or insufficient hardware resources OR adjust # of threads/buffer sizes
  • Improve VDMS ingestion (experiencing LockTimeouts)
  • Improve GPU pipeline to eliminate/minimize tearing in mask frames
  • Reach performance goal of running 1 hr (90% success)
  • Improve CPU pipeline to increase processing rate (if possible)

cwlacewe added 17 commits March 17, 2026 22:14
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: cwlacewe <chaunte.w.lacewell@intel.com>
Signed-off-by: cwlacewe <chaunte.w.lacewell@intel.com>
…pipeline

Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
…point, convert fastapi Dockerfile to multistage, update basis of video clip writer and sending metadata to vdms (WIP)

Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
…ars/capabilities, Update docs, Fix color conversion, decoding, and retry for rtsp, Move model download script and leverage in application, Fix SF test script, [ALL] Format code

Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
…api] Add endpoint for model classes, enable querying, change how video clips are saved (use ffmpeg segment)

Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
…unch.json; [FASTAPI] Update include scripts, update defaiults in default_config, updated test scripts to leverage handler class directly

Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Comment thread finetune/Dockerfile Fixed
cwlacewe added 4 commits June 9, 2026 11:53
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
…g on gpu path, fix

Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
Signed-off-by: Lacewell, Chaunte W <chaunte.w.lacewell@intel.com>
@cwlacewe
cwlacewe marked this pull request as ready for review June 11, 2026 14:42

@s-gobriel s-gobriel left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I see a lot of threads being created, communicating on shared memory and hardware resources are pre-allocated. I don't know if there will be deadlocks but if the demo runs fine then no need to debug all the cases unless there are some issue with the threading. .. Can we track visually the area submitted to object tracking (not objects discovered) .. I want to make sure that for each frame there is some area being sent to YOLO for each frame and no frame is skipped because the threads are blocked ... maybe you can also explain to me how we are queuing the frames and what will happen if this queue is building up in the pipeline

@cwlacewe
cwlacewe merged commit af1f206 into main Jun 11, 2026
4 checks passed
@cwlacewe
cwlacewe deleted the cwl_demo branch June 11, 2026 20:42
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants