Solution for Law enforcement. Investigation of crimes

Accelerated processing and analysis of massive data sets for collection of evidence base on the suspect


When investigating crimes, a large array of data against suspects is collected by technical means of surveillance. It is necessary to process these data with high accuracy and single out only those that testify to the committed crime and will also help identify accomplices.

The main problem is not to miss anything important. It is not so much the processing time that plays a role here, but the accuracy and reliability of data processing. It should be possible to quickly and conveniently adjust the processing results to correct any errors or inaccuracies found.

The main purpose of using Harvester: to speed up the processing of large data sets, filter out unnecessary data, categorize the results by topic, keywords and persons.

Solution description 


All videos are being checked

for the presence of the suspect.


In addition, is possible to analyze photo/video materials in order to detect the necessary objects (for example: money, weapons, prohibited substances, etc.) or people

These can be both materials collected by technical means of surveillance and materials from mobile phones or, for example, the suspect's social networks.


If the files meet the specified criteria

hey are submitted for further processing.


Audio is extracted from video materials

and these files are transcribed together with other collected audio data.


In addition, all text files pass through Sentiment analysis, Summarization & Categorization

by predefined topics.


The investigator receives a categorized database on the specific suspect.

He can then cross-check the information from the text, audio and video files, make sure that the material is suitable for evidence and make appropriate corrections if minor inaccuracies have been made in the processing of the data.

The result of Harvester's work will be the collected evidence base.

Key Technologies 


Face recognition

Object detection

Speaker identification

Speech to text


Keyword spotting


miscellaneous 53 final



Sentiment analysis

Data types

Video & Images

● recognition of a person on the face (identification/verification) and analysis of portrait characteristics● people search● objects search and detection


● recognition of a person by voice (identification/verification)● speech transcription (audio to text conversion)● retraining of language models for transcription● speaker language identification in audio● determining the gender of a person in audio


● language identification● translation of text documents● keyword spotting● sentiment analysis● text document summarization● named entity recognition● text documents● categorization by predefined topics

Try it out with your own data

Need few more text

Thank you!

We will contact you shortly

Can't send form.

Please try again later.

Made with