Recognizing indoor scenes is still regarded an open challenge on the literature. Such scenes can be well represented by their composing objects, which can vary in angle, appearance, besides often being partially occluded. Even though Convolutional Neural Networks are remarkable for image-related problems, the top performances on indoor scenes are from approaches modeling the intricate relationship of objects. Knowing that Recurrent Neural Networks were designed to model
structure from a given sequence, we propose representing an image as a sequence of object-level information in order to feed a bidirectional Long Short-Term Memory network trained for scene classification. We perform a Many-to-Many training approach, such that each element outputs a scene prediction, allowing us to use each prediction to boost recognition. We outperform RNN-based approaches on MIT67, an entirely indoor dataset, while also improving over the most successful methods
through an ensemble of classifiers.

Prof. Erickson Rangel do Nascimento – Advisor (DCC – UFMG)
Prof. Anisio Mendes Lacerda – Co-advisor (DCC – UFMG)
Prof. Wagner Meira Júnior (DCC – UFMG)
Dr. Renato José Martins (Pós Doc DCC – INRIA)