second.pytorch
second.pytorch copied to clipboard
kittiviewer cannot show the detection result when I evaluate testing data in kitti.
I have trained the model and I want to evaluate the testing data in kitti dataset.
I modified the car.fhd.config:
eval_input_reader: {
dataset: {
dataset_class_name: "KittiDataset"
# kitti_info_path: "/root/data/new/second.pytorch/second/KITTI_DATASET_ROOT/kitti_infos_val.pkl"
kitti_info_path: "/root/data/new/second.pytorch/second/KITTI_DATASET_ROOT/kitti_infos_test.pkl"
kitti_root_path: "/root/data/new/second.pytorch/second/KITTI_DATASET_ROOT"
}
And then use the command: " python ./pytorch/train.py evaluate --config_path=./configs/car.fhd.config --model_dir=model_dir/evalmodel1 --measure_time=True --batch_size=1 "
it outputs:
[ 41 1600 1408]
feature_map_size [1, 200, 176]
remain number of infos: 7518
Generate output labels...
[100.0%][===================>][18.55it/s][06:48>00:00]
generate label finished(18.35/s). start eval:
avg example to torch time: 1.279 ms
avg prep time: 4.002 ms
avg voxel_feature_extractor time = 0.171 ms
avg middle forward time = 27.615 ms
avg rpn forward time = 6.117 ms
avg predict time = 14.980 ms
model folder only has one step result. Evaluation succeed or not?
When I try to see the visualization result in kittiviewer, it doesn't have the detection box, what should I do to get the visualization results for evaluation testing data?
Thank you!
I have the same problom,do you solve it?
do you solve it?
I went through it and solved it by making sure that all the paths of the generated files are correct.