KristofferFogh04
commented
3 years ago Hi Dannie125
Thanks for your interest in my project for my master thesis. To start off, at the moment it is very much a work in progress (so it probably should not have been made public just yet), so do not expect it to work out the box at the moment. I expect to polish and finish the repo in the following months, and I also expect to publish the dataset N-AU-DR, which I have collected and used.
The N-AU-DR dataset is a single-class event-based object detection dataset, where the object to detect is drone racing gates. It consists of two parts; a simulated part and a real part. The simulated part is event streams that have been generated using an event camera simulator in AirSim. It is captured from a simulated event camera mounted on a drone flying random trajectories around a visually realistic lab environment where there are drone racing gates. Bounding boxes around the drone racing gates are annotated. The real part is captured using a real event camera mounted on a real drone flying around in a real lab.
I can notify you when I update the repo and publish the dataset. Thanks for the interest.
Best regards Kristoffer
Hi Kristoffer,
Thanks for your sharing the works of Event-based detection. It is very helpful. I know you from uzh-rpg/rpg_asynet/issues:). I am new on it. And I am interested in your "n_au_dr" dataset.
Would you tell me, what is the difference between "Prophesee" and "n_au_dr" ?
And another question: When I tried to run "sparse_RNN", using "Prophesee" instead, and modified some tips, some errors happened. " if self.file_positions[(i_file % self.batch_size) + iself.batch_size + next_start_idx[i_file]][0] != None: IndexError: list index out of range" When I changed this line "self.file_positions = [[None]] (self.total_num_frames + 100000)" to "self.file_positions = [[None]] * (self.total_num_frames + 800000)", another error happened. "if self.file_positions[idx][0] is not None: TypeError: 'NoneType' object is not subscriptable" So I guess some scripts have not been updated to the newest on the gitlab, aren't they?
Thank you.