Inference not predicting on videos

[rpainaik]

Hi! I've been having some issues with inference on predicting videos, specifically with a single instance and centroid model. A certain set of videos that I've been using hasn't been able to have inference run properly on them, and only about 40-50 frames have been labeled after running inference.

My model was trained on 1.3.3 and 1.3.4, but predominantly the latter, and we are using that version of SLEAP to run inference. The videos are all in grayscale with some varying contrast as well as dimensions, but that hasn't posed an issue. I've attached some images from the settings for inference, but please let me know if there's anything else I can add to this.

[eberrigan]

Hi @rpainaik,

Please send some screenshots of your images with the animal and user labels. It's a little hard for me to tell what we are looking for on this image.

Thanks so much!

Elizabeth