DDD17 data format

[ShristiDasBiswas]

Hi,

The paper says that with DDD17, each input comprises 20 voxel grids with 32000 events each.

1. What does nr_bins_per_data=5 in the code mean with respect to the above statement?

2. The print(events.shape) obtained from train_load_sensor is (batch_size, nr_eventsnr_bins_per_data, w, h). So, the paper uses nrevents*nr_bins_per_data number of event frame inputs with only 1 corresponding label for all of them?

[jinglinglingling]

Hi,

The paper says that with DDD17, each input comprises 20 voxel grids with 32000 events each.

1. What does nr_bins_per_data=5 in the code mean with respect to the above statement?
2. The print(events.shape) obtained from train_load_sensor is (batch_size, nr_events_nr_bins_per_data, w, h). So, the paper uses nr_events*nr_bins_per_data number of event frame inputs with only 1 corresponding label for all of them?

Hello, I have the same question, have you solved it

[jinglinglingling]

Hi,

The paper says that with DDD17, each input comprises 20 voxel grids with 32000 events each.

1. What does nr_bins_per_data=5 in the code mean with respect to the above statement?
2. The print(events.shape) obtained from train_load_sensor is (batch_size, nr_events_nr_bins_per_data, w, h). So, the paper uses nr_events*nr_bins_per_data number of event frame inputs with only 1 corresponding label for all of them?

Hi there,
Do you have the ddd17 dataset on the Ev-segnet github page? Because the page for their dataset is no longer accessible