Shorten Production Times and Lower Costs with Machine Learning
PixStor Search with Machine Box automatically generates tags that describe the contents of an asset. That tag can be an actor or other scene components including, location, weather and product placements. An organization can quickly identify relevant existing assets helpful for producing a sequel for example. PixStor Search and Machine Box can process multiple types of data with a high level of accuracy.
Machine Box and PixStor Search Machine Learning
Machine Learning at Scale
Search for codecs, resolutions, project, producer, director, etc. Quickly locate data needed for a project based on the visual and deep-metadata content. Auto-tag data without requiring hours of laborious curation. Know what data you have and ensure nothing is lost.
Automatically Analyze Content to Build a Searchable Index
Machine Box AI analyzes asset content, extracting metadata for input into PixStor Search’s fast full-text search capability to find valuable assets
Quickly Extract Metadata from Assets
With a relatively small set of training data, you can build a model that is able to automatically categorise input data into one or more classes with no need for long training cycles or GPUs
Deep Video Analysis Including Face Detection and Image Classification
Video file frames are extracted, and each frame is processed and the results are collated into a compact and each to use structure
Classify Text, Images, Structured and Unstructured Data
Automatically identify who and what is in an asset, populate metadata enabling searching for specific assets, automatically categorize and group assets
Search by People, Visual Description, or Custom Trained Frame
Identify who is in an image, recognize actors and celebrities using cutting edge search technology
Machine Box Brings Fast, Easy and Accurate Machine Learning to PixStor Search
With Machine Box, PixStor Search harvests metadata directly from images, videos, sequences and documents, reducing the need for human intervention when curating data, significantly reducing costs. Machine Box auto-tagging saves huge amounts of time and money. PixStor Search transcodes data into preview proxies for quick browsing and result validation. PixStor Search provides a user friendly interface for interactively searching across these data sets.
PixStor Search’s Dynamic Guided Search feature aids users to locate the data using the most rapid method of determination possible. Searches can be refined via the easy to use interface, allowing users to drill down to the exact search terms they need, browsing through the data as they go, to help guide the search terms. Shown are results for the search term “night” in a sample dataset.
Runs On-Prem, on Simple Hardware
No need for expensive GPUs or cloud-scale infrastructure, can run offline behind a secure firewall. No data needs to travel to the cloud for computation, but can quickly and temporarily on-demand burst-scale via Kubernetes/AWS to accommodate high scale data processing requirements.
Easily Identify All Assets Across a Filesystem
Searching for needed assets is laborious and error prone. PixStor Search with Machine Box makes search smarter by allowing users to find images based on their content. PixStor Search and Machine Box applies facial recognition, similarity, context and textual search to assets to the reduce search complexity and decrease the chance of missing important data.
PixStor and Machine Box Integration
PixStor and PixStor Search pre-processes files and coordinates with Machine Box on new data ingest, so efficient machine learning and analysis can begin. PixStor Search provides highly efficient file recognition ensuring assets are processed by Machine Box’s Machine Learning at the point of creation or modification of an asset.