search

OAID / Tengine

Tengine is a lite, high performance, modular inference engine for embedded device
Apache License 2.0
4.66k stars 997 forks source link

Using Multiple slice with Tengine mobilenetSSD detector. #199

Open rahulsharma11 opened 4 years ago

rahulsharma11 commented 4 years ago

Hi, I am using a mobilenet SSD detector (91 classes) with tengine. That model contains multiple slice layers (10 slice layers). After checking "dump_graph()", i can see the detection_out layer node shifts from "0" to "10". Is there a different way to use multiple slice layer with Tengine? As it is giving error - cpu_runner.cpp:801 graph: /home/rock64/rahul/mb_lite_coco_caffe_models/deploy_upgraded_withslice.prototxt:4 active num is: 3 output nodes: 2 cpu_runner.cpp:801 graph: /home/rock64/rahul/mb_lite_coco_caffe_models/deploy_upgraded_withslice.prototxt:3 active num is: 3 output nodes: 2 cpu_runner.cpp:801 graph: /home/rock64/rahul/mb_lite_coco_caffe_models/deploy_upgraded_withslice.prototxt:2 active num is: 3 output nodes: 2 cpu_runner.cpp:801 graph: /home/rock64/rahul/mb_lite_coco_caffe_models/deploy_upgraded_withslice.prototxt:1 active num is: 3 output nodes: 2 cpu_runner.cpp:801 graph: /home/rock64/rahul/mb_lite_coco_caffe_models/deploy_upgraded_withslice.prototxt:0 active num is: 3 output nodes: 2

AkkiSony commented 3 years ago

@rahulsharma11 Did you get Mobilenet SSD running? Are you using Khadas VIM3 board?