- The library has been rewritten for this initial release. The library usage is different from before. Refer to the documentation in the README for instructions on requirement installation and usage. Due to an incompatible logging procedure, previous tools need to be used to finish uploading any sequences that were partially uploaded with the previous tools.
- Improved authentication procedure
- Improved user interface with only one executable that can run several basic tools with basic arguments, as well as advanced tools and/or advanced arguments
- Improved logging procedure
- Simpler installation with pip
- Modules can be imported in Python to enable easier development of custom process and/or upload scripts
- Inadequate README #219, #180, #159, #229, #226, #157
- Removal of obsolete scripts #162
- Inadequate logging #63
- Store the subsecond estimations done in sequence processing in case of identical timestamps
- Update the timestamps list in case duplicates are flagged
- Logging was modified to store log files in a hidden directory where the image is located and not in the import path.
- DONE file is created in the import path and not where the tools are being run from.
- exifread version requirement upgraded to resolve installation issues
- support geotagging from a csv file
- support various interpolations
- support re-authentication as a command
- support option to keep the original images intact
- print out version
- increase the amount of information printed out without the verbose flag.
- Create .mapillary folder for image copies in case of passing
--keep_original
- Add missing global variable EPOCH in interpolation
- Do not require filename column to be passed when processing csv, but align csv data and images in order by image file names if filename column is missing
- Support partial matching between image file names and file names in the csv file
- Do not delete the import path in case it is specified as an import path for video files to be sampled into and that particular video file was already sampled into that import path, but rather issue a warning that user needs to delete existing frames that were sampled in any previous runs
- Sample video frames into a made up sub directory
.mapillary/sampled_video_frames/"video filename"
, either where the video is located or in the import path if it is specified. - Add command line arguments for
user_name
,user_email
anduser_password
forauthenticate
command, in order to avoid the prompt.
- Fix bug that resulted in sub seconds added twice in case subseconds were written to capture time tag as well as sub second time tag.
- Additional information and progress bars under verbose in process,
- Partial optimization in DONE file sending,
- Post process command covering process summary, list of image status and options to move image according to status
- Extend support in csv processing
- Adding number of threads and max attempts in upload as command line arguments
- Correct the last interpolated direction
- Enable specification of a time offset in case of already geotagged images
- Add better progress bars and information output in process
- Support Blackvue videos with embedded gps data
- Add a simple
download
command to download all the blurred imaged from Mapillary for a certainimport_path
- Support import of multiple videos
- Argument
--video_file
was renamed to--video_import_path
as directories of one or more videos can now be processed and uploaded. Note that even if one video is to be processed and uploaded, the directory of the video should be specified as the video import path and not the video file itself. - Only the Image Description EXIF tag is overwritten with the mapillary image description which includes all the data obtained during the
process
command. If one would like the rest of the tags to be overwritten, an additional argument needs to be passed under advanced usage. Single specific tags can also be overwritten by additional specific corresponding arguments.