commenting override_base_feature_extractor_hyperparams: true in config, I got
ValueError: SSD Inception V2 feature extractor always usesscope returned by `conv_hyperparams_fn` for both the base feature extractor and the additional layers added since there is no arg_scope defined for the base feature extractor.
and uncommenting the structure is not restored.
@siddharthdas8 by now this strategy is antiquated. I’ve had success converting graphs with tensorflow 2.0 and the tensorrt api in the tensorflow repo: GitHub - tensorflow/tensorrt: TensorFlow/TensorRT integration. I’d try that rather than using a two year old version of the object-detection api.
Hi, I’m a newcomer. Could you elaborate how to solve this problem now?
I don’t know how to commit bash either.
Thank in advance
I have the same problem.
I want to retrain ssd inception v2 model on custom data adn run inference in TensorRT.
If i use object detection api with ae0a9409212d0072938fa60c9f85740bb89ced7e this hash.
is it gonna work??
Which Tensorflow versions should I use??
Currently I am using 1.14.0
Also, What version of Tensorflow should be installed on jetson nano for inference?
Currently I have 1.14.0+nv19.9 on jetson nano.
Please let me know these details.
Thank you
just execute this code ,it will list all the supported operation of your tensorrt x.x.x version
import ctypes
import tensorrt as trt
#ctypes.CDLL(“build/libflattenconcat.so”)
#initialize
TRT_LOGGER = trt.Logger(trt.Logger.INFO)
trt.init_libnvinfer_plugins(TRT_LOGGER, ‘’)
runtime = trt.Runtime(TRT_LOGGER)
PLUGIN_CREATORS = trt.get_plugin_registry().plugin_creator_list
for plugin_creator in PLUGIN_CREATORS:
print(plugin_creator.name)
RnRes2Br1Br2c_TRT
CgPersistentLSTMPlugin_TRT
RnRes2Br2bBr2c_TRT
SingleStepLSTMPlugin
FancyActivation
ResizeNearest
Split
InstanceNormalization
GridAnchor_TRT
NMS_TRT
Reorg_TRT
Region_TRT
Clip_TRT
LReLU_TRT
PriorBox_TRT
Normalize_TRT
RPROI_TRT
BatchedNMS_TRT
FlattenConcat_TRT