Draft
Conversation
…rmats Add load_from_h5() to read iRfcb 0.8.0+ HDF5 classifier output files (.h5) with threshold support. Update load_from_csv() to support use_threshold parameter via the class_name_auto column. File index scanning now discovers *_class*.h5 files alongside CSV and MAT. Loading priority is CSV > H5 > MAT. The classification threshold checkbox now applies to all three formats and is placed below the Classification Folder path in Settings for better discoverability. Includes hdf5r as optional dependency (Suggests), test fixtures, and comprehensive tests for all new functionality.
Add Import PNG → SQLite feature that reads annotations from PNG images organized in class-name subfolders (iRfcb convention). Includes a multi-step UI flow with class mapping dialog for unmatched classes and overwrite warnings for existing samples. New exported functions: scan_png_class_folder() and import_png_folder_to_db(). 32 new tests, vignette and FAQ updates.
Adds filter dropdowns above the class selector in Class Review mode, allowing users to scope reviews by year, month, and IFCB instrument. All filters default to "All". Changing year cascades to update available months and instruments. Database functions list_classes_db and load_class_annotations_db now accept optional filter parameters. New list_annotation_metadata_db function extracts available filter options from annotation sample names.
Connect directly to remote IFCB Dashboard instances to browse samples, download images on demand, and cache them locally. Includes settings UI toggle between local folders and dashboard mode, bulk zip downloads with throttled parallelism, ADC/autoclass on-demand fetching, and graceful SQLite-only fallback when ADC is unavailable for MAT export. New exported functions: parse_dashboard_url(), list_dashboard_bins(), download_dashboard_images(), download_dashboard_images_bulk(), download_dashboard_adc(), download_dashboard_autoclass(), and get_dashboard_cache_dir(). Also fixes reticulate venv binding in test setup.R so scipy tests no longer skip incorrectly.
Add a Predict button in Sample Mode that classifies images using a remote CNN model via iRfcb::ifcb_classify_images(). Gradio API URL and model are configured in Settings > Live Prediction with debounced URL input and dynamic model fetching. Predictions show per-image progress, respect the classification threshold setting, skip manually reclassified images, and auto-add new classes to the class list. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
…ettings Download individual PNGs instead of full zip archives in class review mode, making dashboard class review much faster. Add configurable download parameters (parallel downloads, sleep time, timeout, retries) in Settings. Allow local classification files (CSV/H5/MAT) to be used in dashboard mode as an alternative to dashboard auto-classifications.
Set a 15-second curl timeout (10s connect) on individual image requests so non-existent images fail fast instead of hanging. Stop retrying on 404/410 responses since the image clearly doesn't exist. Track consecutive failures per sample and skip all remaining images from a sample after 2 failures, avoiding long waits when annotations reference samples not available on the dashboard.
Prevent SSRF by rejecting non-HTTP(S) URLs in parse_dashboard_url() and the Gradio predict observer. Fix download_dashboard_autoclass() to parse actual ROI numbers from the CSV pid column instead of assuming sequential 1..N numbering.
This was
linked to
issues
Mar 3, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No description provided.