This file contains pre-trained weights for the UNet backbone of a SplineDist-based object detection model, a variant of StarDist, optimized for detecting Drosophila melanogaster eggs in backlit images of transparent gel substrates. The model was trained for 400 epochs and is designed for use within the counting-3 codebase.To ensure proper functionality, place the model file in the project/models/ directory. The associated configuration file is project/configs/unet_backbone_rand_zoom.json, as referenced in the code.