Can NCSDK2 compile the SSD_model normally?i have some trouble when i compile the frozen_graph.pb to .graph

CChiH13 · ‎11-13-2019

i use ssd_inception_v2 pretrained model to train my own model.

i try to compile it with NCSDK2 and get an error above

Traceback (most recent call last):
  File "/usr/local/bin/mvNCCompile", line 208, in <module>
    args.old_parser, args.cpp, args)
  File "/usr/local/bin/mvNCCompile", line 186, in create_graph
    load_ret = load_network(args, parser, myriad_config)
  File "/usr/local/bin/ncsdk/Controllers/Scheduler.py", line 72, in load_network
    parsedLayers = p.parse(arguments)
  File "/usr/local/bin/ncsdk/Controllers/Parsers/TensorFlow.py", line 659, in parse
    dummy_feed_dict = self._gen_dummy_feed_dict(arguments)
  File "/usr/local/bin/ncsdk/Controllers/Parsers/TensorFlow.py", line 576, in _gen_dummy_feed_dict
    feed_dict = {self.tf_input_op.outputs[0] : np.ones(self.get_input_shape(arguments))}
  File "/usr/local/bin/ncsdk/Controllers/Parsers/TensorFlow.py", line 457, in get_input_shape
    if inferred_input_shape[axis] != None:
IndexError: list index out of range

and than it is my config file i edit it refer this website https://movidius.github.io/ncsdk/tools/tf_ssd_config.html

input_height: 300
input_width: 300
 
postprocessing_params {
    num_classes: 3
    background_label_id: 0
    max_detections: 10
 
    nms_params {
        score_threshold: 0.300000011921
        iou_threshold: 0.600000023842
        max_detections_per_class: 10
    }
 
    box_params {
        var: 0.1
        var: 0.1
        var: 0.2
        var: 0.2
  }
}
 
score_converter: SIGMOID

Sahira_Intel · ‎11-15-2019

Hi Chan,

Did you follow these instructions to compile a Tensorflow network? Can you please provide more information about your model?

I do recommend switching to OpenVINO though, since it is the latest software. It is compatible with the NCS1 and NCS2.

Best,
Sahira