Overview / Usage

Object localization can be defined as the task of finding the bounding boxes of objects in a scene. Most of the state-of-the-art approaches utilize meticulously handcrafted training datasets. The aim of this project is to create a generative adversarial reinforcement learning framework, which can work without having any explicit bounding box information. Instead of relying on bounding boxes, the framework uses tightly cropped object images as training data. Experiments indicate that it is possible to achieve a promising localization performance without having explicit bounding box data. It can be concluded that generative adversarial reinforcement learning is an important tool in dealing with learning problems where explicit input/output paired data is not available.

Methodology / Approach

The object localization framework consists of two parts: a reinforcement learning agent (RL agent) and a discriminator. The RL agent takes input scenes and crops them with the objective of creating a tightly cropped object image. The discriminator tries to distinguish whether the image is generated by the RL agent or it belongs to a tightly cropped object database.

Technologies Used

The project is created using Python and Intel optimized version of TensorFlow. Intel AI DevCloud has been used for simultaneous hyperparameter search and training of the neural networks used in the framework.

Repository

https://github.com/erenhalici/garl_localization

Other links

• ICIP 2018 Paper