Using camera-capture as an annotation tool for using images stored in a file directory

I’ve noticed that the --help response from /usr/local/bin/camera-capture includes an option that looks like its for reading jpgs from a directory as image source:
camera-capture :filie//my_directory/

Is the correct…if so what is the correct syntax for specifying the jpg directory??

Malcolm Capel

We support gstreamer and jetson_multimedia_api. You can run a gstreamer command to read JPEG files and do decoding. Or can try this sample:


Actually we’re not interested in just displaying images we have to annotate them for training an SDD-mobilenet network. In this context annotation consists of establishing a bounding box for relevant objects in the images.



Hi @capel, the camera-capture tool from jetson-inference lacks the UI controls for pausing/seeking through the JPG images, so they would just fly by anyways. Instead, I recommend using the CVAT tool for annotating existing images and exporting them to Pascal VOC format.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.