A single camera/video image may content two or more scripts and the presence of such multiple scripts in a single image makes the text recognition task of the image challenging. Moreover video frames suffer from low resolution, blur, complex background etc. and these factors add extra difficulty in recognition task. Although, there are techniques available for multi-script recognition of scanned document images but only a few pieces of work have been done towards the text recognition of multi-script camera/video images.  In this talk we will discuss about text recognition techniques from multi-script camera/video images. There are two approaches for multi-script text recognition: (i) develop a generalize recognition system which can recognize characters of all scripts (ii) first identify portions of  the  different scripts present in an image and then use script-specific recognition system on the identified portion.  First approach is very difficult and it is, in fact, not possible to develop such a system when number of scripts is to be considered is large. In this talk we will discuss different issues towards these approaches.  For better understanding, demonstrations of some systems will be given to the audience. Finally, some open and new challenging problems will be discussed.