InterDetect EDAPS baseline dataset and model weights

This repository contains the annotations for the Interdetect dataset, as well as parsed images of selected videos and the model weights for the insect detector classifier based on EDAPS. The content of the files is: - Annotations_MicheleGudiati.zip: this archive contains the annotations exports (i.e. text files) as exported from CVAT v1.1 alongside an helper script to parse them into CSV files (CVAT_exports/cvat_anns_parsing.py), and an adapted script to run training using these annotations (train.py). - inter-detect-dataset.zip: this archive contains images used to train the model, as well as parsed annotations in JSON format (dataset_anno_train.json, dataset_anno_val.json). Images are split in training (train/) and validation (val/) sets, with images in each set coming from different videos (but the same overall geographic location and season). - iter_80000.zip: these are the PyTorch model weights, see main github repo for more details.