The ESP32-S3 device cannot correctly identify the exact number of people nearby.

[jackyzp]

I have tested single and multiple ESP32-S3 devices, launching them via Python, Docker, and Rust respectively. Unfortunately, none of the setups can accurately detect the number of people in the surrounding rooms — the system consistently shows a count of 1 person only.
I would like to confirm whether this feature is not actually implemented in the project, or if there are errors in the startup methods and operating procedures. Do we need to import additional models to make it work properly?
I have also tried generating and modifying code with Claude multiple times, but still failed to achieve real-time crowd detection. If anyone has successfully reproduced the actual people counting function, please share the detailed operation steps. Thank you.

[ruvnet]

You need at least 3 esp32 s3, esp32 c6 work better, one isn't going to work.

[jackyzp]

You need at least 3 esp32 s3, esp32 c6 work better, one isn't going to work.你至少需要 3 个 esp32 s3，esp32 c6 效果更好，一个是不行的。

I have used more than three (specifically four or five), yet the results remain the same with no noticeable changes. For instance, when three devices are deployed, only one or two people are detected. However, as people enter or leave, the count never decreases and remains stuck at one or two. I even suspect there must be an issue somewhere, but I still haven't been able to identify the root cause.

[jackyzp]

You need at least 3 esp32 s3, esp32 c6 work better, one isn't going to work.你至少需要 3 个 esp32 s3，esp32 c6 效果更好，一个是不行的。

Do you have any authentic documents or videos that demonstrate how to actually reproduce and real-time detect the number of people around using the ESP32-S3? I want to follow the instructions and give it a try.