Hi @shankarjadhav232, here is the link to Nano SD card image for JetPack 4.4 DP: https://developer.nvidia.com/jetson-nano-sd-card-image-44-dp
give me the link for download jetpack 4.4dp for jetson nano developer kit
give me the link to download deepstream 5.0dp for jetson nano developer kit
Add TensorRT to the list of things jetpack 4.4 brickedā¦
Can you provide information about the issue you are having?
I was trying out one of the featured projects seen hereā¦
When running the model to compile the MobileNet SSD model [compile_ssd_mobilenet.py] i get this.
2020-07-15 10:40:40.450457: I tensorflow/compiler/tf2tensorrt/segment/segment.cc:486] There are 1849 ops of 28 different types in the graph that are not converted to TensorRT: Fill, Merge, Switch, Range, ConcatV2, ZerosLike, Identity, NonMaxSuppressionV3, Squeeze, Mul, ExpandDims, Unpack, TopKV2, Cast, Transpose, Placeholder, Sub, Const, Greater, Shape, Where, Reshape, NoOp, GatherV2, AddV2, Pack, Minimum, StridedSlice, (For more information see Accelerating Inference In TF-TRT User Guide :: NVIDIA Deep Learning Frameworks Documentation).
2020-07-15 10:40:41.872395: I tensorflow/compiler/tf2tensorrt/convert/convert_graph.cc:633] Number of TensorRT candidate segments: 2
2020-07-15 10:40:42.116825: F tensorflow/core/util/device_name_utils.cc:92] Check failed: IsJobName(job)
Aborted (core dumped)
The developer says it has worked on previous versions. Thus another 4.4 issue.
I posted some related topics/workarounds to your GitHub issue here:
https://github.com/dataplayer12/homesecurity/issues/6#issuecomment-658842494
I also saw the way to fix this on JK Jungās blog about JetPack 4.3, and it appears this project hadnāt been updated/tested since JetPack 4.2.
I saw these fixes while googling, but couldnāt see how to work them.
Where is this directory? Canāt find it?
Change tensorrt/tftrt/examples/object_detection/graph_utils.py:98 as follows will fix this core dumped.
98 #node.device = ā/device:CPU:0ā
99 node.device = ā/job:localhost/replica:0/task:0/device:CPU:0ā
Same hereā¦
add this line force_nms_cpu=False in build_detection_graph
frozen_graph, input_names, output_names = build_detection_graph(
config=config_path,
checkpoint=checkpoint_path,
force_nms_cpu=False
)
The project maintainer applied this patch to try in commit 402c76 to their master branch.
please share the documents to run pretrained peoplenet model on deepstream 5.0 with jetpack 4.4
Refer to the Instructions to deploy this model with DeepStream section of this page: https://ngc.nvidia.com/catalog/models/nvidia:tlt_peoplenet
If you have further questions about DeepStream, please post them to the DeepStream forum.
This seems to be a link for jetson NX .will the same work for jetson nano too ?
Sorry about that, the link was mixed-up. Here are the links for JetPack 4.4 DP:
I checked this link there was deepstream 5.0dp is only available for T4 and V100, for jetson nano developer kit deepstream 5.0 was available, please clear out these confusion
Yes, we will update the page to clarify that DeepStream 5.0 is currently on for JetPack 4.4 Developer Preview (L4T R32.4.2). Thanks.
2 Likes
At this point, please post issues you may have in their own topics so we can best address them - thanks. Locking this topic now.