sebastiantombs
Known around here
@kc8tmv The model I've been using is dark.pt. I have also tried the logo.pt model and had terrible results with it, but I was running it with dark.pt and that may be the cause.
From what Ken said about the 100 and 170 errors and the performance of DeepStack outside of the BI arena, I'm guessing that the problems are more related to CPU overload, when using the CPU version especially, and possibly even when running the GPU version because the CPU is still utilized to generate the images for DeepStack while continuing to detect, itself, for the triggers.
From what Ken said about the 100 and 170 errors and the performance of DeepStack outside of the BI arena, I'm guessing that the problems are more related to CPU overload, when using the CPU version especially, and possibly even when running the GPU version because the CPU is still utilized to generate the images for DeepStack while continuing to detect, itself, for the triggers.