Git repo for paper "Enhance embedded object detection system using multi-view"
The baseline experiments were performed using the Ultralytics version of YOLOv5.
train.txt and val.txt contain the training set and the validation set.
Folders exp1_multiview and exp2_camera_position contain the code and test sets needed to perform the multi-view and camera position experiments, respectively. The models to be used are the models trained in the benchmark experiments.
Dataset can be found here.
Visualized results of our baseline experiments and the sweep of hyperparameters can be found here.