Depth-DensePose: an efficient densely connected deep learning model for camera-based localization

Abstract

Camera/image-based localization is important for many emerging applications such as augmented reality (AR), mixed reality, robotics, and self-driving. Camera localization is the problem of estimating both camera position and orientation with respect to an object. Use cases for camera localization depend on two key factors: accuracy and speed (latency). Therefore, this paper proposes Depth-DensePose, an efficient deep learning model for 6-degrees-of-freedom (6-DoF) camera-based localization. The Depth-DensePose utilizes the advantages of both DenseNets and adapted depthwise separable convolution (DS-Conv) to build a deeper and more efficient network. The proposed model consists of iterative depth-dense blocks. Each depth dense block contains two adapted DS-Conv with two kernel sizes 3 and 5, which are useful to retain both low-level as well as high-level features. We evaluate the proposed Depth-DensePose on the Cambridge Landmarks dataset, which shows that the Depth-DensePose outperforms the performance of related deep learning models for camera based localization. Furthermore, extensive experiments were conducted which proven the adapted DS-Conv is more efficient than the standard convolution. Especially, in terms of memory and processing time which is important to real-time and mobile applications.

An Efficient Indoor Localization Based on Deep Attention Learning Model

“…The proposed model was evaluated and compared to related works, such as PoseNet [18], LSTM-Pose [26], DensePoseNet [51], SVS-Pose [29], VLocNet [32], and Depth-DensePose [36]. Table 1 shows our model's quantitative findings and comparisons with similar state-of-the-art methods on the 7-Scenes dataset.…”

“…According to deep learning research, adding more layers makes a model more accurate [36][37][38]. However, issues such as vanishing/exploding features may arise in the deeper model, which would be detrimental to the training outcomes.…”

“…However, issues such as vanishing/exploding features may arise in the deeper model, which would be detrimental to the training outcomes. An efficient deep learning model called Depth-DensePose was proposed in [36] 2 depicts the proposed model's architecture. There are three stages to it.…”

Visual and light detection and ranging-based simultaneous localization and mapping for self-driving cars

“…There are several localization techniques presented in literature that use only cameras [16], [17] or integration of camera [18] to improve performance of these techniques. Suhr et al [19] proposed, in urban environments, using GPS/IMU for global positioning and camera for recognition of road markers as well as lane markers to find both lateral and longitudinal positioning.…”