cbird is a command-line program for finding duplicate images and videos that cannot be found by general methods such as file hashing. Content-Based Image Recognition (CBIR) is used, which examines the pixels of files to get comparable features and "perceptual" hash codes.
The main features are:
- CBIR algorithms for image and video search
- Cached index in the top directory, prevents needless re-indexing
- GUI for evaluating duplicates
- Huge index/database is possible
- Comprehensive file format support
- Search inside zip files
Compile it yourself using my detailed notes.
Add execute permission and run
chmod +x cbird-0.7.0-x86_64.AppImage
./cbird-0.7.0-x86_64.AppImage -install # optional install helper
cbird [...]- Required packages: trash-cli
- Optional packages: ocenaudio, kdenlive
- AppImage won't run (debian, Ubuntu 23.04+)
apt install libfuse2
- Missing libOpenGL.so.0
- debian:
apt install libopengl0 - redhat:
yum install libglvnd-opengl
- debian:
- Missing window titlebar (Fedora)
cbird -platform wayland-egl [...]
- Unzip the distribution file and run
cbird/cbird-mac [...]
- Unzip the distribution file and run the program
- Install helpers (optional): vlc, kdenlive
Optional: create shortcuts for cbird
- Unzip into C:\ so you have C:\cbird\cbird.exe
- Enable script execution
- Run PowerShell as administrator
- Enter
Set-Execution-Policy RemoteSigned - Close PowerShell
- Create profile script (if you don't have one)
- Run PowerShell normally
New-Item -Type File $PROFILE -Force
- Add cbird shortcut to profile
OpenWith $PROFILESet-Alias -Name cbird -Value C:\cbird\cbird.exe
- Shortcut for pictures folder
function cbird-pics {cbird -use $HOME\Pictures $args}
cbird -help is very detailed
cbird -use <path> -update
cbird -update
cbird -dups -show
cbird -similar -show
cbird -p.dht 1 -similar -show
This is lacking documentation at the moment. But for now...
- The GUI is displayed with
-showif there is a selection or results. - GUI windows have a context menu (right click) with all available actions.
- The two deletion actions ("Delete" and "Replace") use the trash/recycler by default. There is no way to permanently delete files (not even with batch deletion commands)
- Find exact duplicates (using file checksums)
- Find modifications
- General transforms: resize, rotate, crop
- Image edits: blur, sharpen, noise, color-grade, grayscale
- Video edits: clipping, fps change, letter boxing
- Evaluate matches
- Compare attributes (resolution, file size, compression ratio)
- Flip between matches
- Zoom-in to see fine details, compression losses
- False-color visualization of differences for fast evaluation
- No-reference subjective quality estimate
- Jpeg compression quality estimate
- Align videos temporally, flip between them or play side-by-side
- File management
- Sort/rename based on similarity
- Rename files using regular expressions
- Move/rename files/directories within the index
Common formats are supported, as well as many obscure formats. The available formats will ultimately vary based on the configuration of Qt and FFmpeg.
cbird -about lists the image and video extensions. Note that video extensions are not checked against FFmpeg at runtime, so they could be unavailable.
Additionally, zip files are supported for images.
To get the most formats you will need to compile FFmpeg and Qt with the necessary options. Additional image formats are also available with kimageformats.
Links are ignored by default. To follow links, use the index option -i.links 1
If the search path contains links, they are only considered when scanning for changes (-update), otherwise there is no special treatment. For example, deleting a link is the same as any other deletion operation.
Duplicate inodes are not followed by default. If there are duplicate inodes in the tree, the first inode in breadth-first traversal is indexed. To follow all inodes, for example to find duplicate hard links, use -i.dups 1.
The index stores relative paths (to the indexed/root path), this makes the index stable if the parent directory changes. However, if a path contains links, or is a link itself, it is stored as-is; which may be less stable than the storing the link target. To store the resolved links instead, use i.resolve 1. This is only possible if the link target is a child of the index root.
Note that cbird does not not prevent broken links from occurring, the link check is temporary during the index update.
The "weed" feature allows fast deletion of deleted files that reappear in the future. A weed record is a pair of file hashes, one is the weed/deleted file, the other is the original/retained file. When the weed shows up again, it can be deleted without inspection (-nuke-weeds)
- Two files are examined (matching pair) -- use
-p.mm 1or-p.eg 1to force pairs - Neither file is a zip member
- When one of the two files is deleted, it is marked as a weed of the first one
There is nothing to prevent deletion of the original/retained file, so the weed record can become invalidated. If the original is no longer present, the association can be unset with the "Forget Weed" command.
cbird -weeds -show # show all weeds
cbird -nuke-weeds # delete all weeds
cbird -similar -with isWeed true # isolate weeds in search results
There are a few for power users.
CBIRD_SETTINGS_FILEoverrides the path to the settings file (cbird -aboutshows the default)CBIRD_TRASH_DIRoverrides the path to trash folder, do not use the system trash binCBIRD_CONSOLE_WIDTHset character width of terminal console (default auto-detect)CBIRD_COLOR_CONSOLEuse colored output even if console says no (default auto-detect)CBIRD_FORCE_COLORSuse colored output even if console is not detectedCBIRD_LOG_TIMESTAMPadd time delta to log messagesCBIRD_NO_BUNDLED_PROGSdo not use bundled programs like ffmpeg in the appimage/binary distributionQT_IMAGE_ALLOC_LIMIT_MBmaximum memory allocation for image files (default 256)QT_SCALE_FACTORglobal scale factor for UITMPDIRoverride default directory for temporary files; used for opening zip file contentsCBIRD_MAXIMIZE_HACKset if window manager/qt is not restoring maximized windows (default auto-detect)
Check the development notes for known bugs and feature ideas.
Report bugs or request features on github
There are several algorithms, some are better than others depending on the situation.
Uses one 64-bit hash per image, similar to pHash. Very fast, good for rescaled images and lightly cropped images.
Uses DCT hashes centered on scale/rotation invariant features, up to 400 per image. Good for heavily cropped images, much faster than ORB.
Uses 256-bit scale/rotation invariant feature descriptors, up to 400 per image. Good for rotated and cropped images, but slow.
Uses histogram of up to 32 colors (256-byte) per image. Sometimes works when all else fails. This is the only algorithm that finds reflected images, others require -p.refl and must rehash the reflected image (very slow)
Uses DCT hashes of video frames. Frames are preprocessed to remove letterboxing. Can also find video thumbnails in the source video since they have the same hash type.
Filters results with a high resolution secondary matcher that finds the exact overlap of an image pair. This is most useful to drop poor matches from fdct and orb. Since it requires decompressing the source/destination image it is extremely slow. It can help to reduce the maximum number of matches per image with -p.mm #
Indexing happens when -update is used. It can take a while the first time, however subsequent updates only consider changes.
Unused algorithms can be disabled to speed up indexing. If you have large images, you may as well enable all algorithms because image decompression dominates the process.
| Arguments | Note | Time (seconds) |
|---|---|---|
| -update | all enabled | 46 |
| -i.algos 0 -update | md5 only | 2 |
| -i.algos 1 -update | +dct | 41 |
| -i.algos 3 -update | +dct features | 44 |
| -i.algos 7 -update | +orb features | 44 |
| -i.algos 15 -update | +color hist | 46 |
Search speed varies with algorithm. The OpenCV search tree for ORB is quite slow compared to others. It is better suited for -similar-to to search a smaller subset suspected to have duplicates.
| Arguments | Note | Time (milliseconds) |
|---|---|---|
| -similar | dct | 54 |
| -p.alg fdct -similar | dct features | 200 |
| -p.alg orb -similar | orb features | 9000 |
| -p.alg color -similar | histograms | 450 |
Indexing large sets of smaller images benefits from disabling algorithms.
| Arguments | Note | Rate (Img/s) | Time (minutes) |
|---|---|---|---|
| -i.algos 0 -update | md5 only | 861 | 9:41 |
| -i.algos 1 -update | +dct | 683 | 12:11 |
| -i.algos 3 -update | +dct features | 377 | 22:04 |
| -i.algos 7 -update | +orb features | 348 | 23:56 |
| -i.algos 15 -update | +colors | 227 | 36:39 |
For N^2 search (-similar) only DCT hash is normally practical.
| Arguments | Note | Time (s) |
|---|---|---|
| -p.dht 1 -similar | dct, threshold 1 | 5.5 |
| -p.dht 2 -similar | dct, threshold 2 | 5.6 |
| -p.dht 3 -similar | dct, threshold 3 | 5.9 |
| -p.dht 4 -similar | dct, threshold 4 | 7.1 |
| -p.dht 5 -similar | dct, threshold 5 | 8.9 |
For K*N (K needle images, N haystack images) the slower algorithms can be practical even for large datasets. For a quick test we can select and search for the first 10 items:
cbird -p.alg fdct -select-type i -head 10 -similar-to @
| Arguments | Note | Time (s) |
|---|---|---|
| -p.alg dct -p.dht 2 | dct, threshold 2 | 1.3 |
| -p.alg fdct -p.dht 7 | dct-features, threshold 7 | 1.5 |
| -p.alg orb | orb-features | 84.41 |
| -p.alg color | colors | dnf2 |
- Mac OS X port and x86 binary package
- Use current libraries (FFmpeg, Qt6, Quazip)
- Command Line
- describe what -filter is doing
- cleanup some logging
- Indexing
- Use "nice"/low priority when scanning/indexing
- 3-4x faster jpeg indexing using idct scaling
- Recognize more video extensions
- Remove colon (":") workaround for powershell
- Failsafe cache updates using write-temporary-move method
- Fix slow cache update on network volumes
- Fix -i.ignored to actually track all ignored stuff
- Searching
- Add % progress for sql queries, cleanup logging
- Search using a given hash code
- Use numeric sort for filenames (e.g. 1.jpg sorts before 10.jpg)
- Functions are always chainable and consistent between commands
- Properties: add aspectRatio
- Disable inode check on win32, too slow on network volumes
- Fix attempt to template match video result
- Viewing
- Add theme system, remove hardcoded theme/style
- auto-detect dark/light theme on mac/windows
- shade background windows when modal window is displayed
- "-theme" to change the theme
- Streamline context menu
- ffplay-sbs: tons of improvements
- Do not require index for -select-files
- DPI-aware image scaling
- Show creation time metadata for videos too
- Browse the selected item in its folder
- Use grid for folder view and fixed thumbnail aspect ratio
- Cycle contents thumbnails when hovering in folder view
- Add index/folder thumbnail tool
- Fix some performance issues
- Fix keyboard navigation in folder view
- Fix missing match range
- Prevent crash/hang on some invalid inputs (corrupt videos etc)
- Fix slow qualityscore on network volumes
- Fix potential crash on two-tone images
- Fix potential crash/hang with corrupt video files
- Fix elided strings display
- Add theme system, remove hardcoded theme/style
- Add weeds feature (recommended by r/user/traal)
- Search: Add max-threshold
-p.mtwhich produces a result for every needle until threshold is exceeded - Indexer: Use skip-loop-filter to speedup video indexing ~20%
- Indexer: Option to show all ignored files
- Indexer: Handle links sanely by default, add options
i.links,-i.dups,-i.resolvefor other cases. - Filters: Add -head/-tail
- Image view: difference image can align to template match
- Video view: fix issues with unequal frame rates
- Video view: export the aligned pair to Kdenlive
- Speedups: video compare view, differences image,
-group-by