Accomplished
I started this week by reviewing the design report feedback and working on incorporating it into my squat-rack design. I also got the squat-rack setup and detection workflow working on the Jetson using offline video. This gives me a working starting point on our intended hardware. I was able to achieve the same model accuracy as the laptop deployment, which was expected, but it was a good sanity check.
The next step is to test the workflow with frames arriving continuously from a camera. I still need to establish whether live inference keeps up with our intended checking rate. I will also profile memory usage, latency, and throughput so we have actual measurements to use when planning the combined workload.
Schedule status
The offline Jetson run completes part of the hardware work planned in my last report. However, profiling remains unfinished, and live-camera testing and integration are still behind our earlier schedule. We have not yet run all three workflows together on one Jetson.
To catch up, I will prioritize the live rack test and individual performance measurements next week, then work with Jerry and Alan on a combined run. Measuring each workflow separately will give us a baseline for identifying any slowdown or memory pressure when they run together.
Deliverables for next week
- Run the squat-rack workflow with a live camera feed on the Jetson.
- Measure memory usage, per-check latency, and throughput, including whether live processing falls behind incoming frames.
- Continue incorporating the design report feedback into the rack workflow and testing plan.
- Work with Jerry and Alan to run all three workflows together on one Jetson and measure their combined performance.