Luis Tobias

Thanks for your answer @lu-wang-g . I have added to my target the following dependencies: deps = [ "@com_google_absl//absl/flags:flag", "@com_google_absl//absl/flags:parse", "@com_google_absl//absl/status", "@com_google_absl//absl/strings", "@com_google_absl//absl/strings:str_format", "@org_tensorflow_lite_support//tensorflow_lite_support/cc/port:statusor", "@org_tensorflow_lite_support//tensorflow_lite_support/cc/task/core:external_file_handler", "@org_tensorflow_lite_support//tensorflow_lite_support/cc/task/core/proto:external_file_proto_inc", "@org_tensorflow_lite_support//tensorflow_lite_support/cc/task/vision:object_detector", "@org_tensorflow_lite_support//tensorflow_lite_support/cc/task/vision/proto:bounding_box_proto_inc", "@org_tensorflow_lite_support//tensorflow_lite_support/cc/task/vision/proto:class_proto_inc", "@org_tensorflow_lite_support//tensorflow_lite_support/cc/task/vision/proto:detections_proto_inc",...

Dusty, I figured it out, it turns that the values for ``` ${jetson-inference_LIBS}, ${jetson-utils_LIBS}, ${jetson-inference_INCLUDE_DIRS} and ${jetson-utils_INCLUDE_DIRS} ``` are retruning empty. Once i hardcoded the paths to the .so and...

Thanks @dusty-nv this was indeed the solution. Best Luis.

I believe the way they extract the features is the key. In the paper it is mentioned: features are extracted with an overlap of 8 frames, then averaged across the...

In order to make it work, the preprocessing has to be corrected: for (uint32_t i = 0; i < input.size(); ++i) Otherwise you will run into an infinite Loop