Our goal is to be able to detect objects dynamically with the camera on the robot.
See requirements.txt
for Python dependencies.
Additionally, install Protobuf for your system and run:
protoc object_detection/protos/*.proto --python_out=.
To install protobuf, go here, download the appropriate zip for your OS (not language necessarily), unzip it to a folder, and set your PATH to also point to the bin.
Later, in order to avoid an import error, run the following line (must do for each new terminal opened): export LD_PRELOAD=/usr/lib/aarch64-linux-gnu/libgomp.so.1
and activate the virtual environment with the following line: source venv/bin/activate
Tensorflow's research models for a frozen Object Detection model (SSD Mobilenet V1 COCO, retrieved from previous commit)
Tensorflow's research models for their object_detection
library found under research/
in models.
Tensorflow's research models for the labels provided by the object_detection
library.