Detecting and identifying object in real time is one of the crucial way for the human’s to interact with the environment. In addition, it is also followed by taking some actions or performing reaction to those objects. Object detection in computer vision deals with analyzing all the parts and different scales of the image to make sure all the object instances have been identified. When it comes to real time object detection things get even more complicated as all the analysis needs to be done in milliseconds just like human system performs identification. This presented thesis work aims at achieving similar type of performance for object detection in real time which will be used in the car interior along with performing various actions related to detected objects as part of intelligent interior in the autonomous car. Another challenging part here is to perform the defined actions for an object without any time latency i.e. no delay in performing actions. The algorithm used here to perform real time object detection, works at 18-19 fps on a system with NVIDIA GTX 965M (4GB memory) graphics card which of course, increases with more powerful graphics card. For transferring the detected data like confidence, bounding box coordinates and object class etc, it is first encapsulated into a ’JSON’ object for convenient parsing at target client application. This JSON object is then transmitted using TCP/IP and information is extracted at client application. And based on the extracted information, different actions are performed inside the autonomous car. YOLOv3, the real time object detection algorithm used in this thesis is modified to improve the fps on the NVIDIA GTX 965M (4GB memory) graphics card and also, it is made capable of acting as a server and connect to target client applications in order to transfer real time data.