Harvester
Compliance Monitoring Software
" Designed to scan and detect content across networks and the internet to help satisfy any corporate compliancy policy. "
Download Datasheet - 381kb
Features
- High Speed Scanning
- Visual reporting
- Analytical reporting
- Scalability
- Similis image/video fingerprint plug-in
- ICA pornographic detection plug-in
- Accepts 3rd party plug-ins
- Small Footprint
- API available for third-party integration
How it Works
Harvester Compliance software is an application designed to automatically retrieve information in the form of image and video content from internet web sites and news groups, as well as from standalone and networked PCs and servers. The data scanned is passed to a number of plug-in filters that automatically classify the content. The results are reported back in customizable analytical formats for forensic review by the person or persons using the software.
Harvester is a highly effective tool for use in detecting pre fingerprinted image and video file content as well as dynamically detecting inappropriate image content. It is ideally suited as a Compliance Monitoring software management tool for use on corporate and institutional networks, social networking web sites and in other communities which host large volumes of user generated content. This forensic software tool is aimed at content owners, enterprises and website owners to help identify content types.
Harvester's image mapping methodology accurately matches exact and closely matched frames for optimum reliability. Typically snippets of a video lasting only 15 seconds can be successfully matched against original files in this way. Detailed reporting of web sites scanned and files matched/not matched are generated and exported to data sheets for forensic confirmation.
Web crawling detection can be improved by using these reports to monitor content reappearing on detected URLs.
Software Plugins to Harvester Software
Similis Image Fingerprinting
The Similis image plug-in to Harvester Compliance software provides a method of fingerprinting image content that is typically found across company and institutional networks. Although other fingerprinting systems exist for document detection and MD5# matching, Similis specifically fingerprints images providing a number of distinct advantages that can substantially increase effectiveness when there is a need to keep track of what is being stored on a network and by whom. When an image is fingerprinted by the Similis system not only will the original image be detected but so will versions of the same image that have been altered, edited or resaved in different image file formats.
Using Harvester software with the Similis plug-in is easy. After running a network scan selected images can be quickly fingerprinted and the next time a scan is carried out the software will detect the fingerprinted images but also images that have been edited and altered in some way.
Altered Images
Images are commonly altered in a number of ways to escape detection from MD5# filtering. Such methods include: Resizing, file format change, color changing, flipping, mirroring, rotation, cropping addition of text, turning to monochrome, turning to negative. Depending on the degree of change, Similis is successfully able to detect fingerprinted images altered in these ways.
Accuracy
Independent testing confirms high levels of accuracy for matching altered images details of which can be supplied upon request.
Similis Image Usage Case
In the digital world in which we now live it is difficult to keep up to date with the different types of content being circulated and stored by end users. Unacceptable images may include subjects such as terror, hate, and bullying, defamatory, libelous or controversial religions. One of the issues we face is how to determine if media content is unacceptable and how companies and institutions really know what is actually stored on their networks. Similis gives an organisation a tool to enable it to collect information about the media stored on its network and where it is located. Once the information has been collected into an easy to read report, the system administrator can then fingerprint, track and enforce company rules on its employees. Similis lets the organization decide what content is either acceptable or unacceptable by providing a quick, easy and cost effective way of empowering system administrators to perform the task without costly overheads while substantially reduce risk to an organization.
Similis Video Fingerprinting
The Similis Video plug-in to Harvester Compliance software provides a similar solution to the Similis image plug-in but is designed to handle video formats which involve different file handling protocols. Similis video ingests video frames at a speed of up to 40 times faster than real time to create a fingerprint for subsequent matching of scanned files. This means a one hour video could be ingested in just over one minute. The size of the fingerprint depends on the size of the file but on average this would be 24KBs which avoids latency in look ups that can be typical of database solutions. The matching of video files against the fingerprint is fast averaging speeds 20 times faster than real time.
Accuracy
Accuracy in correctly identifying video content is higher than image fingerprinting as there are more images within the frames to match. Typically snippets of a video lasting only 20 seconds can be successfully matched against original files. Extensive independent testing confirms high accuracy levels, details of which can be provided upon request.
Similis Video Usage Case
With the event of video sharing websites like YouTube employers are very likely to find video content that breaches copyright law or acceptability standards on the organization’s business computer and network system. It is hard for the system administrator to find time to review all content regularly to determine if it is appropriate, legal or authorized for consumption. Similis video provides a tool that can find the video files on the network and place them in a report that makes it easy and time effective for the system administrator to assess the content and subsequently fingerprint for detection, filtering or removal. Additionally recurring images found in other locations on a network can be detected and deleted and new instances of the same image entering the network can be similarly detected. In the same way as Similis image works Similis video detects altered and edited content for example if it has been included in ‘mash up’.
ICA Detection of Pornographic Content
ICA (Image Composition Analysis)
The ICA plug-in, real time Image Composition Analysis, is the most widely used image analysis software for the detection of inappropriate pornographic image content. In percentage terms the software has been independently tested to provide accuracy levels in excess of ninety percent.
The risk of managing internet content has grown considerably. Responsible for costing organizations millions in non-business related internet activity, which can also lead to damaged business reputations, harassment cases and employees claims.
Video Scanning for pornographic content reduces the need for human monitoring of video content. Designed to accurately scan video files to detect and categorise sensitive content, the software is a practical solution for the essential maintenance and supervision of content. This unique frame-by-frame analysis effortlessly detects illicit and adult material.
Integration to Existing Platforms and Workflows
Harvester Compliancy software and its associated plug-ins have small footprints of less than 2MB and can easily plug-in to existing automated review systems. Harvester and associated plug-ins can be supplied as a total software system or as APIs with DLLs to enable easy integration into customers’ existing work flow architecture. The scanning results are reported in XML and score values can be fed to existing automated systems to provide coordinated return values. Alternatively Harvester comes with its own customizable GUI as shown in the evaluation software or may be used by customers in its present form. Harvester operates effectively at all levels.
Supported Operating Systems
The Harvester software application has been designed to operate cross platform, supporting Windows, Linux and UNIX operating systems.
Minimum System Requirements
- Standard Server/PC: 300 Mhz or higher Processor (Compatible processors such as Pentium II, Pentium III and AMD Processors are also supported)
- Memory: 32 MB RAM (64MB Recommended) Hard Disc: 10 Mb Hard Disc Space
- Drive: CD-ROM drive