Code for 3D object detection for autonomous driving
MIT License
933
stars
349
forks
source link
You need to create a preprocessing config, such as those in mb_preprocessing that includes the classes Car Pedestrian and Cyclist. After that you need to set up preprocessing in gen_mini_batches.py to process All using the config you set up. If you want more classes you will at least also have to modify kitti_dataset.py as it is hardcoded to name joint detection of Pedestrian and Cyclist as 'People' and Pedestrian, Cyclist and Car as 'All', but I haven't tested expanding this yet. #157
You need to create a preprocessing config, such as those in mb_preprocessing that includes the classes Car Pedestrian and Cyclist. After that you need to set up preprocessing in gen_mini_batches.py to process All using the config you set up. If you want more classes you will at least also have to modify kitti_dataset.py as it is hardcoded to name joint detection of Pedestrian and Cyclist as 'People' and Pedestrian, Cyclist and Car as 'All', but I haven't tested expanding this yet.
Keep in mind that results on pedestrians and cyclists will not be very good out of the box, likely due to the poor balancing of the classes in the dataset.
Hi! I create multi_class trainning before gen_min_batch with car people and cyclists.
when i val this model, only one class (car) is showing in result!
I wodder if you have same issue,please tell me why ?
You need to create a preprocessing config, such as those in mb_preprocessing that includes the classes Car Pedestrian and Cyclist. After that you need to set up preprocessing in gen_mini_batches.py to process All using the config you set up. If you want more classes you will at least also have to modify kitti_dataset.py as it is hardcoded to name joint detection of Pedestrian and Cyclist as 'People' and Pedestrian, Cyclist and Car as 'All', but I haven't tested expanding this yet.
I have forked the repo and done some work of my own, you can see examples of the two first files I mentioned at https://github.com/Fredrik00/avod/blob/master/avod/configs/mb_preprocessing/test_all.config and https://github.com/Fredrik00/avod/blob/master/scripts/preprocessing/gen_mini_batches.py
Keep in mind that results on pedestrians and cyclists will not be very good out of the box, likely due to the poor balancing of the classes in the dataset.
Originally posted by @Fredrik00 in https://github.com/kujason/avod/issues/124#issuecomment-469181349
Hi! I create multi_class trainning before gen_min_batch with car people and cyclists. when i val this model, only one class (car) is showing in result! I wodder if you have same issue,please tell me why ?