My Projects for the specialization of : TensorFlow Advanced Techniques Specialization from deeplearning.ai
-
Course 1 : Knowing how to use and customize the TF API.
- Building multi output model.
- Using custom layers / models / losses / optimizers / callbacks.
- Implementing those custom parameters from scratch.
- Building Vgg / resnet from scratch.
-
Course 2 : Applying advanced TF functionalities for effective custom-training processes.
- Gradient tape usage.
- Applying custom training loops.
- Ising Autograph to use the graph-style code instead of Eager Excution.
- Applying the different strategies for training, depending on the resourcess available.
-
Course 3 : Advanced Computer vision techniques related to state of the art models and tasks.
- Creating an Object localization model from scratch.
- Using the Object Detection API for inference.
- Pretraining and using Retina-net for object detection.
- Building U-net / VGG - FCN-8 encoder-decoder / Mask R-CNN models from scratch and using them for image segmentation.
- Interpreting the CNNs and how to visualize the performance of each layer in the network using various methods.
-
Course 4 : Applying custom-training techniques to build complex Generative models.
- Building Neural Style Transfere from scratch.
- Building AutoEncoders and Deep AutoEncoders.
- Building Convolutional AutoEncoders from scratch and Applying it to several Datasets.
- Building Variational AutoEncoders from scratch and Applying it to several Datasets.
- Using VAEs to generate Anime Faces. -needed more like another day of training tho-
- Building Vanilla-GAN and DCGAN from scratch.
- Using DCGANs to generate hand sign images.