New version : v3.0.0-beta.3

Releasing v3.0.0-beta.3 🎉️ , featuring:

• Complete rewrite of Docker setup / install process in order to use docker-compose
• Update darknet fork once again using the new -dontdraw_bbox flag AlexeyAB kindly added for us : https://github.com/AlexeyAB/darknet/commit/da050367a8b174f751a2c9720195d51a369c4a4a
• Expose tracker config IOU and UNMATCHED FRAME tolerance
• Check if config.json version matches Opendatacam version
• Clean up code for drag& drop feature… maybe add some UI tweaks like a drop indicator and a specific progress message
• Clean up and improve counter logic code
• Update API docs with new counters attributes
• Add Threshold to filter out large detections
• Update OpenMojis

How to test / install

If you were using OpenDataCam before with docker, the setup has changed to use docker-compose now for several reasons

• simplify the install process
• provide flexibility for people to add other docker services around opendatacam
• enable people to deploy on IoT cloud platform like Balena
• makes the workflow with Docker lighter, I think to deploy a new version of OpenDataCam it will be much fast now as we don’t need to push everything, the base image will not change… also maybe open the door to build from the cloud
• be independant from jetpack version issues … (right now we have problem with darknet and jetpack 4.4)

We still have a “one command” install process but you need to install docker-compose first, please read: https://github.com/opendatacam/opendatacam#1-software-pre-requisite-

Note that right now the raspberry pi cam isn’t working on the Jetson nano using this new setup… we are trying to fix this… and you might run into other issues as this is a significant change from how things worked previously

Still TODO

• @fabianmoronzirfas added support for env variables: https://github.com/opendatacam/opendatacam/pull/222
• Add counter settings “minAngleWithCountingLine” : https://github.com/opendatacam/opendatacam/blob/development/documentation/CONFIG.md#counter-settings
• Update documentation CONFIG.md as many default changed now… we run on the demo.mp4 file now… and file support is via drag & drop to simplify install process
• Update documentation Jetson nano
• Try to fix problem with RaspberryPi Cam on jetson nano not working with new docker setup: https://github.com/opendatacam/opendatacam/issues/178
• Document how to run from custom weights with docker (possible now)
• Document gstreamer use and pipelines
• Update Documentation for yolov3-tiny-prn and yolov4 weights
• Add a “ethics” message
• Change picture of OpenDataCam with something with the jetson nano and not the tx2 ?

[Denizzje]

Better late than never 😉.

• Docker compose install works flawlessly.
• I am on the development branch I think for the non docker install and it works perfect on Jetpack 4.4 as that is where my Xavier NX came with. So you either fixed the issues or it is Nano only.

[jvlies]

Yes I can execute v4l2-ctl. I made an error with my previous post. I did not use ‘0’ but ‘-c’.And that works, as I indicated.Now running couple of days 12 hours non-stop. On Tuesday, 23 June 2020, 20:45:53 CEST, Thibault Durand [e-mail] wrote:

@jvlies if you try this command does it work ? v4l2-ctl --list-formats-ext ?? Maybe you need to sudo apt-get install -y v4l-utils

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or unsubscribe.