Homography Errors

[jgeller112]

Hi!

So I was able to get the offlineInterface to detect files for a single participant/pair of glasses

Finding worldview and gaze data...
Scanning files...: 100%|██████████████████████████████████████████████████████████████████████████| 17/17 [00:00<00:00, 9559.35it/s]
Found 1 device recordings
Found 1 worldview videos
Found 1 gaze csv files
Found 1 world csv files
Found 1 fixations csv files
Found 1 blinks csv files
Found 1 events csv files
Found 1 imu csv files
Found 1 saccades csv files
Found 1 3d_eye_states csv files

When I try tohomography I run into a bunch of errors that I hope you can help me troubleshoot.

Finding worldview and gaze data...
Scanning files...: 100%|█████████████████████████████████████████████████████████████████████████| 17/17 [00:00<00:00, 13025.79it/s]
Found 1 device recordings
Found 1 worldview videos
Found 1 gaze csv files
Found 1 world csv files
Found 1 fixations csv files
Found 1 blinks csv files
Found 1 events csv files
Found 1 imu csv files
Found 1 saccades csv files
Found 1 3d_eye_states csv files
? What action would you like to perform ? Perform Homography
? Use offset-corrected timestamps for homography? No
? Would you like to add a search key for filtering glasses file paths? No
? Enable multi-threading for concurrent processing? (Disable if you're not sure about available compute resources) Yes
? Input path of the CentralCam recording dir. /Users/jasongeller/SocialEyes/src/centralCam/outputs/SEC_Lecture6_102825/
? Select path for homography results (transformed gaze in CentralCam coordinates) /Users/jasongeller/SocialEyes/src/centralCam/outpu
ts/
Scanning files...: 100%|█████████████████████████████████████████████████████████████████████████| 17/17 [00:00<00:00, 21600.48it/s]
Starting homography for 1 recordings
Using device cpu
Loaded SuperPoint model
Loaded SuperGlue model ("outdoor" weights)
Traceback (most recent call last):
  File "/Users/jasongeller/SocialEyes/src/offlineInterface/main.py", line 265, in <module>
    init_homography(session.root_dir, session.cam_dir, output_dir=session.hom_op_dir, multi_thread=multi_thread, search_key = search_key, offset_corrected = offset_corrected)
  File "/Users/jasongeller/SocialEyes/src/offlineInterface/../homography/main.py", line 121, in init_homography
    result = future.result()
  File "/Users/jasongeller/anaconda3/lib/python3.10/concurrent/futures/_base.py", line 451, in result
    return self.__get_result()
  File "/Users/jasongeller/anaconda3/lib/python3.10/concurrent/futures/_base.py", line 403, in __get_result
    raise self._exception
  File "/Users/jasongeller/anaconda3/lib/python3.10/concurrent/futures/thread.py", line 58, in run
    result = self.fn(*self.args, **self.kwargs)
  File "/Users/jasongeller/SocialEyes/src/offlineInterface/../homography/homography_processor.py", line 92, in perform_homography
    self.init_dataframes()
  File "/Users/jasongeller/SocialEyes/src/offlineInterface/../homography/homography_processor.py", line 125, in init_dataframes
    self.world_timestamps_df = CSVProcessor(self.glasses_timestamp_path, oftype, ['timestamp_corrected']).read_csv()
  File "/Users/jasongeller/SocialEyes/src/offlineInterface/../offlineInterface/csv_processor.py", line 39, in read_csv
    data = pd.read_csv(self.file_path, dtype=self.dtype,
  File "/Users/jasongeller/anaconda3/lib/python3.10/site-packages/pandas/io/parsers/readers.py", line 1026, in read_csv
    return _read(filepath_or_buffer, kwds)
  File "/Users/jasongeller/anaconda3/lib/python3.10/site-packages/pandas/io/parsers/readers.py", line 620, in _read
    parser = TextFileReader(filepath_or_buffer, **kwds)
  File "/Users/jasongeller/anaconda3/lib/python3.10/site-packages/pandas/io/parsers/readers.py", line 1620, in __init__
    self._engine = self._make_engine(f, self.engine)
  File "/Users/jasongeller/anaconda3/lib/python3.10/site-packages/pandas/io/parsers/readers.py", line 1898, in _make_engine
    return mapping[engine](f, **self.options)
  File "/Users/jasongeller/anaconda3/lib/python3.10/site-packages/pandas/io/parsers/c_parser_wrapper.py", line 140, in __init__
    self._validate_usecols_names(usecols, self.orig_names)
  File "/Users/jasongeller/anaconda3/lib/python3.10/site-packages/pandas/io/parsers/base_parser.py", line 988, in _validate_usecols_names
    raise ValueError(
ValueError: Usecols do not match columns, columns expected but not found: ['timestamp_corrected']

Any ideas?

I ran offset correction first.

[jgeller112]

In both cases (offset corrected and not offset) it is expecting the timestamp_corrected column. However, this is not a file that can be selected when running the offset module.

[ShreshthSaxena]

Hi @jgeller112, it seems like the offset correction was not successful in that case. You should see new csv files (for the selected datastreams) with a prefix of "ts_corr_" and they should have a "timestamp_corrected" column. We did push some changes so that the device naming scheme (IP address vs. G???) doesn't interefere with offset correction so please try pulling the recent version and running offset correction again.

[jgeller112]

I think it might be an issue in homography and the files it wants?

[jgeller112]

The central timestamp file is contained in the central camera folder. This does not have the corrected timestamp. The homography module is expecting this to have the corrected timestamp. However, This path is not in the config file.

[ShreshthSaxena]

Yes, the central camera issue has been resolved in our recent push. We've been using arducam in our previous collections so the standard method was not compatible with the homography module because of different column types. On your centralCam output csv, if you just rename the "timestamp [ns]" column to "corrected_timestamp" then homography should work fine. For future data collection or piloting, please pull the recent changes and that should resolve it.

[jgeller112]

Good to know! I'll download the most recent version and give it a go tomorrow!

[jgeller112]

Processing now!