YoloV8 Custom problem
-
we have a model trained in YoloV8, starting from Yolo8M, we did all the conversion process according to the documentation, starting from an ONNX, we did everything following the documentation but at the time of running the model we get this:
Jul 09 18:41:59 m0054 bash[4052]: Fault thread: voxl-tflite-ser(tid: 6348)
Jul 09 18:41:59 m0054 bash[4052]: Fault address: (nil)
Jul 09 18:41:59 m0054 bash[4052]: Unknown reason.
Jul 09 18:41:59 m0054 systemd[1]: voxl-tflite-server.service: Main process exited, code=killed, status=11/SEGV
Jul 09 18:41:59 m0054 systemd[1]: voxl-tflite-server.service: Failed with result 'signal'.
Jul 09 18:41:59 m0054 systemd[1]: voxl-tflite-server.service: Service hold-off time over, scheduling restart.
Jul 09 18:41:59 m0054 systemd[1]: voxl-tflite-server.service: Scheduled restart job, restart counter is at 53.
Jul 09 18:41:59 m0054 systemd[1]: Stopped voxl-tflite-server.
Jul 09 18:41:59 m0054 systemd[1]: Started voxl-tflite-server.
Jul 09 18:42:00 m0054 bash[6590]: INFO: Created TensorFlow Lite delegate for GPU.
Jul 09 18:42:12 m0054 bash[6590]: INFO: Initialized OpenCL-based API.
Jul 09 18:42:12 m0054 bash[6590]: INFO: Created 1 GPU delegate kernels.
Jul 09 18:42:12 m0054 bash[6590]: ------VOXL TFLite Server------
Jul 09 18:42:12 m0054 bash[6590]: Segmentation fault:
Jul 09 18:42:12 m0054 bash[6590]: Fault thread: voxl-tflite-ser(tid: 8954)
Jul 09 18:42:12 m0054 bash[6590]: Fault address: (nil)
Jul 09 18:42:12 m0054 bash[6590]: Unknown reason.
Jul 09 18:42:12 m0054 systemd[1]: voxl-tflite-server.service: Main process exited, code=killed, status=11/SEGV
Jul 09 18:42:12 m0054 systemd[1]: voxl-tflite-server.service: Failed with result 'signal'.
Jul 09 18:42:12 m0054 systemd[1]: voxl-tflite-server.service: Service hold-off time over, scheduling restart.
Jul 09 18:42:12 m0054 systemd[1]: voxl-tflite-server.service: Scheduled restart job, restart counter is at 54.
Jul 09 18:42:12 m0054 systemd[1]: Stopped voxl-tflite-server.
Jul 09 18:42:12 m0054 systemd[1]: Started voxl-tflite-server.
Jul 09 18:42:13 m0054 bash[9299]: INFO: Created TensorFlow Lite delegate for GPU.
Jul 09 18:42:24 m0054 bash[9299]: INFO: Initialized OpenCL-based API.
Jul 09 18:42:24 m0054 bash[9299]: INFO: Created 1 GPU delegate kernels.we have tried with the quantification proposed in the documentation and nothing, we are stuck with this, what can we do?
voxl-suite: 1.4.5
p.s. we also found an error in the voxl-configure-tfliteserver code, where this was shown when setting the custom model: set_param_bool expected 2 args, we corrected it this way: set_param_bool requires_labels true
