The significant power of deep learning networks has led to enormous development in object detection. Over the last few years, object detector frameworks have achieved tremendous success in both accuracy and efficiency. However, their ability is far from that of human beings due to several factors, occlusion being one of them. Since occlusion can happen in various locations, scale, and ratio, it is very difficult to handle. In this paper, we address the challenges in occlusion handling in generic object detection in both outdoor and indoor scenes, then we refer to the recent works that have been carried out to overcome these challenges. Finally, we discuss some possible future directions of research.

Although current computer vision systems are closer to the human intelligence when it comes to comprehending the visible world than previously, their performance is hindered when objects are partially occluded. Since we live in a dynamic and complex environment, we encounter more occluded objects than fully visible ones. Therefore, instilling the capability of amodal perception into those vision systems is crucial. However, overcoming occlusion is difficult and comes with its own challenges. The generative adversarial network (GAN), on the other hand, is renowned for its generative power in producing data from a random noise distribution that approaches the samples that come from real data distributions. In this survey, we outline the existing works wherein GAN is utilized in addressing the challenges of overcoming occlusion, namely amodal segmentation, amodal content completion, order recovery, and acquiring training data. We provide a summary of the type of GAN, loss function, the dataset, and the results of each work. We present an overview of the implemented GAN architectures in various applications of amodal completion. We also discuss the common objective functions that are applied in training GAN for occlusion-handling tasks. Lastly, we discuss several open issues and potential future directions.

Nowadays steganography plays an important role in information security. There are various techniques to apply steganography; one such technique is Least Significant Bit (LSB). Due to consecutive embedding of data, LSB is highly vulnerable to data retrieval and manipulation. This study proposes a mixture between Rubik’s cube principle to scramble the audio data, and a modified LSB technique to hide the secret data. The modified LSB technique includes embedding using only irredundant bits of binary representation of each character in the secret message, and hiding in the lowest sample between two consecutive samples of the cover audio. The used technique makes the retrieval of secret message harder because it adds two levels of protection (scrambling, and hiding in the lowest sample) against the attempts of obtaining data; and makes the embedded data imperceptible.

Predicting the full appearance of a partially visible object, known as amodal content completion, remains a challenging task. Many existing approaches struggle to recover missing regions without leaving artifacts from the occluding objects or generating out of context content. To address this limitation, we introduce a two-stage framework that incorporates a weighted mask together with multiple layers of enhanced gated convolution. This design allows the model to focus more on features extracted from the visible region of the occluded object and to propagate its effect through the network. We also integrate weighted global contextual information with these extracted features, which the model uses to refine the initial output produced by the model's first stage. We evaluate our approach on the COCOA dataset and two of its subsets. The experimental results demonstrate that the weighted mask effectively helps the model prioritize the visible areas of partially occluded objects, leading to reconstructions with richer semantic details and more contextually accurate content than the baseline models, particularly for objects with uniform textures such as animals.