Created by: peerajak

Hi I just use VOC on ssd, using standard config file ssd_vgg16_300_voc.yml After saving the model, it produce error in init.cc file as follow.

I trained VOC with YOLOv3, and it works fine without this following error

2020-09-22 14:15:23,490-INFO: iter: 1000, lr: 0.001000, 'loss': '11.070778', time: 0.223, eta: 7:22:16 2020-09-22 14:15:23,511-INFO: Save model to output/ssd_vgg16_300_vocmy/1000. W0922 14:15:28.514690 4997 init.cc:226] Warning: PaddlePaddle catches a failure signal, it may not work properly W0922 14:15:28.514734 4997 init.cc:228] You could check whether you killed PaddlePaddle thread/process accidentally or report the case to PaddlePaddle W0922 14:15:28.514755 4997 init.cc:231] The detail failure signal is:

W0922 14:15:28.514765 4997 init.cc:234] *** Aborted at 1600758928 (unix time) try "date -d @1600758928" if you are using GNU date *** W0922 14:15:28.518110 4997 init.cc:234] PC: @ 0x0 (unknown) W0922 14:15:28.519321 4997 init.cc:234] *** SIGSEGV (@0x0) received by PID 4796 (TID 0x7f2d0dfff700) from PID 0; stack trace: *** W0922 14:15:28.520992 4997 init.cc:234] @ 0x7f2f93acff20 (unknown) W0922 14:15:28.522526 4997 init.cc:234] @ 0x0 (unknown) ./train_esp32plate.sh: line 3: 4796 Segmentation fault (core dumped) python -u tools/train.py -c configs/ssd/ssd_vgg16_300_vocmy.yml --use_vdl=True --vdl_log_dir=vdl_vocesp32plate_dir/scalar --eval