High-Performance Open-Source Archive
ifcb_extract_features(), which computes the slim
feature set (version 4) and blob masks from raw IFCB data by calling the
WHOI ifcb-features
Python package. Features (<bin>_features_v4.csv) and
blobs (<bin>_blobs_v4.zip) are written to separate,
user-specified folders, existing outputs are skipped unless
overwrite = TRUE, and bins can be processed in parallel via
parallel = TRUE / n_cores. A cli
progress bar advances as each bin is processed (for both sequential and
parallel runs), and interrupting the function (e.g. ESC / Stop) reliably
terminates the parallel worker processes instead of leaving them writing
files in the background.features and features_ref arguments
to ifcb_py_install() to optionally install the WHOI
ifcb-features package (and its dependencies) from GitHub,
as required by ifcb_extract_features(). By default the
latest published release is installed; features_ref can pin
a specific tag or install the development branch. When installing into
an existing virtual environment, the install is skipped if
ifcb-features already imports successfully (unless
features_ref is supplied), avoiding a slow repeated
download.dataset_name argument to
ifcb_list_dashboard_bins() to optionally restrict the
listing to bins from a specific dataset. This argument remains useful
for self-hosted dashboard instances that have not yet updated to remove
the api/list_bins endpoint.IRFCB_PYTHON_VENV environment
variable. When USE_IRFCB_PYTHON = "TRUE", you can now set
IRFCB_PYTHON_VENV to either a named virtualenv or a full
path to a venv directory to control which Python environment is
activated on package load. If unset, the previous behavior of
auto-discovering a venv named iRfcb is retained.ifcb_list_dashboard_bins() is deprecated. The upstream
IFCB Dashboard removed the api/list_bins endpoint on
2026-03-08 (WHOIGit/ifcbdb@8c5839f1),
so the function no longer works against the WHOI dashboard and other
deployments tracking upstream. Use
ifcb_download_dashboard_metadata() instead, which retrieves
the same per-bin information from the still-supported
api/export_metadata endpoint.ifcb_py_install() are now installed with a
clean dependency resolution (no longer using pip
--ignore-installed). Previously, installing packages with
pinned, compiled dependencies (such as
ifcb-features/pyifcb, which pin exact
numpy/scipy/pandas versions)
could layer incompatible builds on top of existing ones and corrupt the
environment
(e.g. ImportError: cannot import name '_spropack').gradio_url for
ifcb_classify_images(),
ifcb_classify_sample(),
ifcb_classify_models(), and
ifcb_save_classification() has changed from the Hugging
Face example Space (https://irfcb-classify.hf.space) to a
more stable instance hosted on SciLifeLab Serve
(https://ifcb.serve.scilifelab.se). The default
model_name has correspondingly been updated to
"SMHI NIVA SYKE SAMS SZN ResNet 50 V6". The Hugging Face
Space remains documented as a free alternative for testing and
demonstration.stop(),
warning(), message()) and
utils::txtProgressBar to the cli package.
Errors, warnings, and informational messages now use semantic inline
markup (file paths, argument names, function names, values) and
pluralization. Progress bars are rendered via
cli::cli_progress_bar(). cli is now an
Imports dependency.adc_get_roi_columns() failing to detect ROI
dimension columns for older IFCB instruments (e.g. IFCB110) where HDR
files use different column name casing
(ROIwidth/ROIheight/start_byte)
compared to newer instruments
(RoiWidth/RoiHeight/StartByte).
Column matching is now case-insensitive and positional fallback uses
column indices instead of V-prefixed names (#77).ifcb_classify_images() to classify one or
more pre-extracted IFCB PNG images through a CNN model served by a
Gradio application, returning a data frame of predicted class names and
confidence scores. Per-class thresholds are applied automatically.ifcb_classify_sample() to classify all
images in a raw IFCB sample (.roi file) without prior PNG
extraction. Internally extracts images to a temporary directory and
delegates to ifcb_classify_images().ifcb_save_classification() to classify
IFCB samples via Gradio API and save results as HDF5 (.h5),
MAT (.mat), or CSV (.csv) files.ifcb_classify_models() to list available
CNN models from a Gradio classification server..h5) and CSV (.csv)
classification file support to ifcb_extract_biovolumes(),
ifcb_extract_classified_images(),
ifcb_summarize_class_counts(),
ifcb_summarize_biovolumes(), and
summarize_TBclass(), in addition to existing
.mat support.ifcb_extract_pngs(),
ifcb_extract_annotated_images(), and
ifcb_extract_classified_images()) now preserve raw pixel
values by default (normalize = FALSE), producing images
comparable to IFCB Dashboard and other standard IFCB software.
Previously, pixel values were stretched to the full 0-255 range using
min-max normalization. This change can affect classifier training
results. Set normalize = TRUE to restore the previous
behavior (#75).ifcb_create_manual_file() now writes
class2use_auto as a numeric matrix, matching the format
produced by ifcb-analysis (#74).micron_factor in
ifcb_psd() and ifcb_extract_biovolumes().skip_class in
ifcb_extract_annotated_images().ifcb_run_image_gallery() is deprecated in favor of
ClassiPyR::run_app(). See https://europeanifcbgroup.github.io/ClassiPyR/ for more
information.old_adc in ifcb_extract_pngs(),
ifcb_extract_annotated_images(), and
ifcb_extract_classified_images(). ADC format (old IFCB1-6
vs new) is now auto-detected from the HDR file’s
ADCFileFormat parameter and the ADC column count.mat_files in ifcb_extract_biovolumes() and
ifcb_summarize_biovolumes() (replaced by
class_files).mat_recursive in ifcb_extract_biovolumes()
and ifcb_summarize_biovolumes() (replaced by
class_recursive).ifcb_annotate_samples() to create manual
classification .mat files compatible with the
ifcb-analysis MATLAB repository, using PNG images organized
in class named subfolders and a class2use.mat file.ifcb_zip_images_by_class() to zip each PNG
subfolder with optional random sampling. Useful for preparing
class-specific image archives for submission.diatom_include argument to
ifcb_extract_biovolumes() and ifcb_is_diatom()
for manually forcing specific taxa to be treated as diatoms (overrides
WoRMS classification).timestamp_param argument to
ifcb_get_ferrybox_data() allowing the Ferrybox timestamp
column to be specified dynamically instead of being hard coded.max_time_diff_min argument to
ifcb_get_ferrybox_data() controlling the maximum allowed
time difference in minutes when matching Ferrybox data to requested
timestamps.biovolume_only argument to
ifcb_read_features() to allow reading only biovolume
related columns, improving performance for large feature tables.add_trailing_numbers argument to
ifcb_extract_annotated_images() to control whether a
zero-padded numeric suffix based on the manual class index is appended
to class names in the output filenames.include_classes argument to
ifcb_prepare_whoi_plankton() to allow explicit selection of
classes to include during processing.\donttest{}
instead of \dontrun{}.ifcb_get_ferrybox_data() is now
more flexible and can fall back to the closest available Ferrybox
observation within the specified time window when no exact or rounded
match is found.ifcb_summarize_biovolumes() and
ifcb_extract_biovolumes() are now more flexible and accept
individual .mat files in addition to folders.ifcb_extract_biovolumes() and
ifcb_summarize_biovolumes().ifcb_get_ecotaxa_example().ifcb_create_manual_file() and
ifcb_create_empty_manual_file() now correctly handles
NaN values in the classlist.ifcb_create_empty_manual_file() has been renamed to
ifcb_create_manual_file().ifcb_match_taxa_names() is now superseded by
SHARK4R::match_worms_taxa().mat_folder in ifcb_summarize_biovolumes()
and ifcb_extract_biovolumes() (replaced by
mat_files).expected_checksum in
ifcb_download_test_data().ifcb_download_dashboard_metadata() and
ifcb_list_dashboard_bins().diatom_include parameter to
ifcb_extract_biovolumes() and
ifcb_summarize_biovolumes() for manually forcing specific
taxa to be treated as diatoms (overrides WoRMS classification)
(#65).bins parameter to ifcb_psd() for
selecting which bins to process.fea_v parameter to ifcb_psd() for
selecting feature-file version.use_plot_subfolders parameter to
ifcb_psd() to optionally save plots in subdirectories of
plot_folder based on flag status.flags parameter to ifcb_psd_plot()
to optionally add the quality flag annotation to the plot.ifcb_extract_biovolumes() now support both filename
formats _fea_v*.csv and _features_v*.csv,
increasing compatibility with legacy and new output formats (#61).ifcb_read_features(),
ifcb_summarize_png_metadata(),
ifcb_summarize_biovolumes(), and
ifcb_extract_biovolumes() now include an optional parameter
to select specific feature file versions (e.g., _v2,
_v4), allowing finer control over which feature data are
read and processed.$data, $fits and $flags
data frames returned by ifcb_psd() now use full bin names
(<sample>_<ifcb>) as sample names, improving
uniqueness and consistency with downstream workflows.$data and $fits data frames returned
by ifcb_psd() now preserves the original column names,
including names starting with numbers or containing special
characters.ifcb_psd() has
been replaced by u in $data headers.$flags headers in ifcb_psd() to
use sample instead of file, ensuring
consistent naming across all outputs.ifcb_psd() when plot_folder is specified,
improving processing speed.ifcb_download_test_data() gains checksum validation,
keep_zip option, and improved retry logic.ifcb_extract_biovolumes() and
ifcb_summarize_biovolumes() gain a
drop_zero_volume option to exclude artifacts with zero
biovolumeifcb_read_features() and
ifcb_extract_biovolumes() now handles single
feature_files correctly.ifcb_download_dashboard_data()
and ifcb_download_whoi_plankton().NEWS.md to track package changes.ifcb_download_dashboard_data() for fetching data from
the WHOI IFCB Dashboard.ifcb_download_whoi_plankton() for downloading
WHOI-Plankton PNG datasets.ifcb_prepare_whoi_plankton() to process WHOI data for
integration.vignette("whoi-plankton-data-integration").USE_IRFCB_PYTHON environment variable when
loading the package.ifcb_extract_pngs()ifcb_extract_classified_images()ifcb_extract_annotated_images()gamma argument to
ifcb_extract_annotated_images () and
ifcb_extract_classified_images() for gamma correction.ifcb_is_near_land():
plot = TRUE.source.utm_zone, which is now
determined automatically from longitude.ifcb_read_hdr_data() now correctly handles files from
IFCB Acquire 1.x.x.x (#41).ifcb_convert_filenames() is more efficient and now
correctly parses filenames like "IFCB1_2010_309_192918"
(#40).ifcb_py_install() has changed to
“~/.virtualenvs/iRfcb”..adc file format (IFCB1-6) by
argument old_adc in:
ifcb_extract_pngs()ifcb_extract_classified_images()ifcb_extract_annotated_images()ifcb_read_mat() now returns consistent data structures
aligned with R.matlab::readMat() (#50).ifcb_summarize_png_data() is now defunct (previously
deprecated in version 0.3.11).adc_folder in ifcb_annotate_batch()
(replaced by adc_files).unclassified_id in
ifcb_create_empty_manual_file() (replaced by
classlist).utm_zone in ifcb_is_near_land() (now
determined automatically from longitude).ifcb_create_empty_manual_file() now accepts a complete
class list via classlist, replacing the older
unclassified_id.ifcb_summarize_biovolumes() now handles custom class
lists.fs and
shinytest.imager (replaced by png) in
ifcb_extract_pngs() and base64enc
dependencies.gamma argument to
ifcb_extract_pngs().verbose argument to functions:
ifcb_download_test_data()ifcb_extract_biovolumes()ifcb_is_diatom()ifcb_read_features()ifcb_summarize_biovolumes()iRfcb:::retrieve_worms_records()) to top-level function:
ifcb_match_taxa_names().ifcb_get_ferrybox_data() where multiple
close ferrybox timestamps caused duplicate rows. Now returns only the
nearest match.ifcb_get_shark_example() and
ifcb_get_shark_colnames() for testing and
documentation.ifcb_merge_manual() to merge manual
.mat datasets.ifcb_adjust_classes() to adjust classes in manual
data.ifcb_create_class2use() to generate a class2use
file.ifcb_create_empty_manual_file() to create new
manual .mat files.ifcb_annotate_batch() to annotate multiple
.mat files in a batch based on .png
images..mat files to save space using
do_compression argument in:
ifcb_correct_annotation()ifcb_replace_mat_values()ifcb_summarize_png_metadata() to summarize
EcoTaxa metadata.ifcb_get_ecotaxa_example() to extract EcoTaxa
headers and example data.iRfcb:::split_large_zip() in
ifcb_zip_pngs().ifcb_summarize_png_data() (replaced by
ifcb_summarize_png_counts()).ifcb_get_svea_position().iRfcb:::handle_missing_positions().manual_folder,
feature_folder, and class_folder in several
functions (ifcb_count_mat_annotations(),
ifcb_extract_biovolumes(),
ifcb_read_features(),
ifcb_summarize_biovolumes().nrow(taxa_list) == 0 in
ifcb_count_mat_annotations().mat_recursive option to
ifcb_count_mat_annotations().manual_recursive option to
ifcb_extract_annotated_images() and
ifcb_zip_matlab().roi_recursive option to
ifcb_extract_annotated_images().data_recursive option to
ifcb_zip_matlab().feature_recursive option to
ifcb_zip_matlab().marine_only to ifcb_is_diatom(),
iRfcb:::retrieve_worms_records(),
ifcb_summarize_biovolumes() and
ifcb_extract_biovolumes().feature_recursive to
ifcb_extract_biovolumes() and
ifcb_summarize_biovolumes().mat_recursive to
ifcb_extract_biovolumes() and
ifcb_summarize_biovolumes().hdr_recursive to
ifcb_summarize_biovolumes().ifcb_is_diatom().ifcb_summarize_biovolumes().ifcb_is_in_basin().ifcb_get_shark_example() and
ifcb_get_shark_colnames().iRfcb:::retrieve_worms_records().ifcb_get_svea_position() (replaced by
ifcb_get_ferrybox_data()).ifcb_summarize_biovolumes().sleep_time parameter for
ifcb_download_test_data()..mat count edge case.curl for downloads.ifcb_psd() with
argument micron_factor.testthat for improved
stability.verbose argument to several functions to provide
detailed progress messages during execution.ifcb_replace_mat_values() now only handles
.mat files in the manual_folder.ifcb_is_near_land() now returns NA if
coordinates passed to the function contain NA values.ifcb_get_trophic_type() to assign trophic
strategy to taxa.ifcb_get_shark_colnames():
WADEP, PDMET,
METFP, IFCBNO, TRPHY,
ABUND, and BIOVOL.SAMPLE_TIME,
ABUND_UNITS_PER_LITER, BIOVOL_PER_SAMPLE,
BIOVOL_PER_LITER, C_CONC_PER_LITER, and
SEA_BASIN.extract_aphia_id(): Extract AphiaID from WoRMS
record.extract_class(): Extract taxonomic class from WoRMS
record.handle_missing_positions(): Handle missing positions by
rounding timestamps.ifcb_extract_biovolumes(): Compute biovolumes and
carbon from IFCB data.ifcb_get_shark_colnames(): Retrieve column names for
SHARK submission.ifcb_get_svea_position(): Extract GPS coordinates from
ferrybox data.ifcb_is_diatom(): Identify diatoms in a taxa list.ifcb_is_in_basin(): Check whether points fall inside a
sea basin.ifcb_psd_plot(): Create particle size distribution
plots from IFCB data.ifcb_read_features(): Read IFCB feature files from a
specified folder.ifcb_summarize_biovolumes(): Summarize biovolumes and
carbon content.ifcb_summarize_class_counts(): Count TreeBagger
classifier outputs.ifcb_which_basin(): Return name of sea basin a point
belongs to.summarize_TBclass(): Summarize TreeBagger classifier
results.vol2C_lgdiatom(): Convert biovolume to carbon for large
diatoms.vol2C_nondiatom(): Convert biovolume to carbon for
non-diatom protists.ifcb_read_hdr_data() where
gps_only filtering could fail.ifcb_volume_analyzed() when
inhibition time == 0 (#2).ifcb_is_near_land() by applying
sf::st_wrap_dateline() only when the CRS is
geographic.Initial development release of iRfcb.
ifcb_convert_filenames()ifcb_correct_annotation()ifcb_count_mat_annotations()ifcb_create_manifest()ifcb_download_test_data()ifcb_extract_annotated_images()ifcb_extract_classified_images()ifcb_extract_pngs()ifcb_get_mat_names()ifcb_get_mat_variables()ifcb_get_runtime()ifcb_is_near_land()ifcb_psd()ifcb_py_install()ifcb_read_hdr_data()ifcb_read_summary()ifcb_replace_mat_values()ifcb_run_image_gallery()ifcb_summarize_png_data()ifcb_volume_analyzed_from_adc()ifcb_volume_analyzed()ifcb_zip_matlab()ifcb_zip_pngs()
Need mirroring services?
Contact our team at info@vpspulse.com.
Mirror powered by VPSpulse
Infrastructure sponsored by VPSPulse & Secure Payments by ArionPay.