Created by: ghimiredhikura

Hi,

All the steps (caffe->paddle via X2Paddle->model_optimize_too) are okay, but finally testing with paddle-lite object detection demo it gives segmentation fault. Please help. Here I attached original caffe-model and converted files.

I used this script to convert caffe to paddle-lite .

wget https://github.com/PaddlePaddle/Paddle-Lite/releases/download/v2.0.0/model_optimize_tool
chmod +x model_optimize_tool
x2paddle --framework=caffe --prototxt=my_ssd.prototxt --weight=my_ssd.caffemodel --save_dir=pd_model 
./model_optimize_tool --model_dir=pd_model/inference_model/ --optimize_out_type=naive_buffer --optimize_out=paddle_lite_my_ssd --valid_targets=arm

And used following script for testing object detection https://github.com/PaddlePaddle/Paddle-Lite-Demo/tree/master/PaddleLite-armlinux-demo/object_detection_demo

paddle_lite_my_ssd.zip caffe_my_ssd.zip

Error: Segmentation fault. While tracing the error it comes from this line of the object detection code: https://github.com/PaddlePaddle/Paddle-Lite-Demo/blob/master/PaddleLite-armlinux-demo/object_detection_demo/object_detection_demo.cc#L252