About
Videoindex scans your video file library, identifying all objects within the videos. This allows you to subsequently locate specific video segments simply by typing the name of an object into the search bar.This feature is particularly useful for video editing workflows—enabling you to quickly find necessary footage—or simply for locating a specific file you wish to watch.
To begin using Videoindex, navigate to the "Indexation" window; specify a folder containing your video files, or add individual files for scanning, then click the "Start" button.
Once the indexing process is complete, you will see a checkmark next to each added folder, indicating that all files within that directory have been successfully indexed.
To perform a search, enter a keyword into the "Search" window; results will appear dynamically as you type.You can locate the actual files within your file system by clicking the "Reveal File" button; additionally, you can instantly play back any file, starting specifically from the point where the searched object was detected.The entire list of found files can be exported in XML format for subsequent import into a video editing application—resulting in a timeline sequence where all relevant clips are automatically arranged, cued to the exact moments where the objects appear.This process saves you a tremendous amount of time that would otherwise be spent manually sifting through raw footage to find specific shots!
Utilizing Core ML Machine Learning Models
Videoindex employs trained machine learning models in the Core ML format to identify objects within your videos.The application ships with two pre-installed models: YOLOv3 and ResNet.You can use these models either individually or in combination; furthermore, the application supports the use of your own custom Core ML model files.To manage your models, open the "Settings" window, where three slots are available for loading models.The first slot is dedicated to the YOLOv3 model, which supports the display of "Bounding Boxes" (visual outlines) around recognized objects during the scanning process; the second slot is reserved for the ResNet model; and the third slot is available for loading your own custom Core ML model files.
Automatic Monitoring of Your Video Library
Videoindex continuously monitors the directories you have added for scanning, keeping track of any changes or updates to your video library.In other words, if you add or remove files from a designated folder at any point, these changes will be recorded in the Videoindex database upon the next launch. If new files are detected, the status of the scanned folder will update to indicate that the directory requires rescanning, allowing you to initiate the process of scanning the new video files.
Background Scanning
Depending on the size and quantity of your video files, the scanning process may take a considerable amount of time. Therefore, the application includes an option to hide the Videoindex windows while the program is running; this can be done via a menu item located in the macOS desktop menu bar. Additionally, you can configure the application to launch automatically upon system startup.To enable this feature, select the corresponding option within the Videoindex settings window.
Recommendations for Optimizing Videoindex Performance
Different Core ML models operate at varying speeds.For faster scanning, we recommend using only the Resnet model; to do this, uncheck the box next to the YOLOv3 model in the settings and check the box next to the Resnet model.We also recommend setting the frame scanning interval to 3 seconds.
You have the option to adjust the object recognition confidence threshold—the minimum percentage required for detected objects to be added to the database. Lowering this threshold helps reduce the overall size of the database, thereby increasing the speed of subsequent object searches.
Popup Slide Context
1. Add a folder containing your video files.
2. Select either the built-in Core ML recognition models or use your own custom models.
3. Click the "Start" button to begin the scanning process.
4. Afterward, use the "Search" window to locate specific objects within your videos.5. You can export the detected video segments as an XML file for import into video editing software, or simply locate the corresponding files directly in Finder.