Modify for event_integration #2
Conversation
| } | ||
| if (event_frame_msg.data[pos] < 255) // prevent overflow | ||
| event_frame_msg.data[pos]++; // feed this into model | ||
| // event_frame_msg.data[pos] = 255 // use this for display purpose (visualize) |
There was a problem hiding this comment.
@aamini use this line if you want to visualize the event input at the same time. This doesn't influence the event+RGB model, but it may mess up the old event model that takes in normalized event_frames rather than one-hot event_frames.
| { | ||
| event_array_msg->header.stamp = e.ts; | ||
| } | ||
| if (event_frame_msg.data[pos] < 255) // prevent overflow |
There was a problem hiding this comment.
Should there be an else statement with some warning in case of Overflow happening.
| int e_x = caerPolarityEventGetX(event); | ||
| int e_y = caerPolarityEventGetY(event); | ||
| bool e_polarity = caerPolarityEventGetPolarity(event); | ||
| int pos = e_y * event_frame_msg.step + e_x * 3 + e_polarity; |
There was a problem hiding this comment.
Maybe move the variable declarations outside of the loop to the beginning of the method? Not sure if it will make a big difference as the compiler should be smart enough to optimize.
Considering the amount of work they did with DAVIS, I believe they should have thought about this. Their dvs_renderer package (https://github.com/mit-drl/rpg_dvs_ros/tree/master/dvs_renderer) subscribes to events and output integrated frames ready for display. However:
|
No description provided.