(a) It is required to develop an object detection system with input images of size (128-by-128-by-3) that will produce the bounding box and the classification for 5 different object categories. Explain, Sketch and Discuss (i) the training data preparation, (ii) the full-structure of the network used, (iii) the training steps Using the pre-trained AlexNet shown below and taking the features from (FC8) layer. (3) Input data Convi Conv2 Conv3 Convi Conv5 FC6 FC? FCR 13x 13304 1313x30 13x13256 27x27x256 55x55x96 2272273 1000 (b) Explain and Describe the steps required to design and train an image Coloring (input is a gray image and output is the colored image) system using Pix-2-Pix GAN neural network. Sketch the network, and describe the training data required, the structure of the network and how it is trained. (3)