nnstreamer

nnstreamer copied to clipboard

Cannot open tensorflow2-lite-custom(/usr/lib/nnstreamer/filters/libnnstreamer_filter_tensorflow2-lite-custom.so)

2

```bash $ nnstreamer-check NNStreamer version: 2.4.0 loaded : TRUE path : /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libnnstreamer.so NNStreamer configuration: ============================================================ Configuration Loaded: TRUE Configuration file path: /etc/nnstreamer.ini Candidates: envvar(NNSTREAMER_CONF): (null) build-config: /etc/nnstreamer.ini hard-coded: /etc/nnstreamer.ini [Common]...

wooksong

Update contributor/committer list and beautify it.

1

- Beautify it with https://github.com/all-contributors/all-contributors - Update the list in .md and reviewer list - Held TSC to update members.

myungjoo

Filter::edgetpu tflite forward compatibility broken

1

#4194 Step 1. Create compiler errors for newer versions Step 2. Make it compatible

myungjoo

result string from ml-agent

1

TBU, string from ml-agent interfaces (get model or resource) ``` model_get_all: [ {"version":"1","active":"F","path":"/opt/usr/shared/my_model_file_1.pb","description":"","app_info":""}, {"version":"2","active":"F","path":"/opt/usr/shared/my_model_file_2.pb","description":"","app_info":""}, {"version":"3","active":"T","path":"/opt/usr/shared/my_model_file_3.pb","description":"","app_info":""} ] model_get: {"version":"2","active":"F","path":"/opt/usr/shared/my_model_file_2.pb","description":"","app_info":""} model_get_activated: {"version":"3","active":"T","path":"/opt/usr/shared/my_model_file_3.pb","description":"","app_info":""} resource_get: [ {"path":"/opt/usr/shared/my_resource_file1.dat","description":"","app_info":""}, {"path":"/opt/usr/shared/my_resource_file2.dat","description":"","app_info":""}, {"path":"/opt/usr/shared/my_resource_file3.dat","description":"","app_info":""} ] ```

jaeyun-jung

Device MLOPS Requirements

3

1. Model Deployment, - On-device training support! (Tizen 7.5 M2) - Model deployment (Tizen 7.5 M1): clients (2 groups in g, 1 group in b, 1 group in Tizen). -...

myungjoo

[ML Service] Version control of Model and ML Pipeline

1

## Functional Requirement * Store the previous version if users want (Default: Update the value) * Get the latest version for specific key * Support to rollback to the specific...

again4you

ExecuTorch filter subplugin & Tizen porting

5

PyTorch is releasing ExecuTorch for edge device inferences. - [ ] Tizen porting (tizen-next) - [ ] tensor-fitler::ExecuTorch (in C++) - [ ] test cases - [ ] sample applications...

myungjoo

tensor-transform: better exception handling for incorrect caps/operators

1

Refer: [#4102 ](https://github.com/nnstreamer/nnstreamer/issues/4102#issuecomment-1570862323) If the given operator/options are not compatible with the given gst-caps, don't just throw: ``` ** (tutorial2:5602): CRITICAL **: 16:18:24.653: _check_tensors: assertion 'config->info.num_tensors >= limit' failed 0:00:00.404629212...

myungjoo

Inconsistency between tensor configurations in tensor_filter

2

If we want to use a flexible tensor in tensor_filter, then the gstreamer pipeline will put this information into the capabilities, that, for the tensor_filter, are stored in GstTensorFilter.sinkpad/srcpad.get_current_caps.get_structure.GstTensorConfig.tensor_format. In...

bruecks

How to write custom tensor filter with different size of output tensor from frame to frame

3

I would like to write a custom tensor filter that accepts a static tensor and returns a flexible tensor. The dimensions of the flexible tensor are different from frame to...

bruecks