Features In Different Views Research Articles

Joint contrastive self-supervised learning and weak-orthogonal product quantization for fast image retrieval

Hashing and quantization methods based on deep learning are being widely used in large-scale image retrieval. Most methods use traditional supervised models to train models. Although these methods achieve excellent performance, they rely too much on expensive label information and do not fully utilize data resources. These methods overly encourage binary codes to retain original information, which may lead to models building too much useless information at the expense of discriminative information that is more important for retrieval. To address these issues, we propose a self-supervised method that does not require labeled information, called Contrastive Self-Supervised Weak-Orthogonal Product Quantization (CSWPQ). The method is trained in a label-free self-supervised manner to reduce the need for labels by maximizing the similarity of different views of the same image and minimizing the similarity of different image views, which not only improves the generalization ability of the model but also enhances the robustness. We introduce a quantified contrastive learning method that is different from traditional contrastive learning to learn more discriminative image representations by constructing quantified image features of different views of the image. In addition, in order to reduce the quantization error caused by product quantization, we impose a weak-orthogonal constraint to increase the diversity of representations, reduce the information loss and computational overhead in the quantization process, and improve the computational speed. Extensive experiments on CIFAR-10, NUS-WIDE and FLICKR25K datasets show that our proposed method achieves better performance.

Multi-view clustering via double spaces structure learning and adaptive multiple projection regression learning

Multi-view clustering aims to group objects with high similarity into one group according to the heterogeneous features of different views. The graph-based clustering methods have obtained excellent results. However, there remain a few common drawbacks. For example, some methods do not consider graphs' high-order structure information. Thus, fuller data information cannot be obtained. In addition, some methods remove noise, outliers, and redundant information in the graph learning phase, resulting in the loss of graph information. Furthermore, using predefined graphs cannot exploit complementary information between views. A triple strategy-based multi-view clustering method is presented to solve the above issues. First, Laplacian graphs are used for fusion learning, and the underlying first-order and second-order structure information among views are explored simultaneously. Then, a label fusion scheme is designed to eliminate noise, outliers, and redundant information and to mine the intrinsic characteristics of data labels. Besides, the consistent label matrix in adaptive regression learning is used to explore complementary information between views in a mutually guided learning way. Finally, the objective function is solved by using an efficient iterative method. Six types of experiments are conducted on eleven real-world multi-view datasets, and the conclusions that can be drawn are: (1) the proposed algorithm achieves the best results in terms of clustering accuracy on ten datasets with an average accuracy improvement of 5.11% compared to other algorithms. Specifically, the accuracy improved by 9.05% on dataset HW and 10.95% on dataset Reuters compared to the second results; (2) The ablation experiments confirm that the different learning strategies included in the proposed algorithm allow it to achieve better clustering performance.

Multi-view human pose and shape estimation via mesh-aligned voxel interpolation

Although multi-view human pose and shape regression methods have information from other views for complementing and correcting, existing ones still have its own drawback of not fully taking advantage of multi-view setup. Thus they are far from efficiently aligning and merging features in different views. In order to tackle these problems, we propose a multi-view framework where features from all views are well aligned and merged through multi-view voxel aggregation with inverse projection. Our framework highlights three major characteristics. Firstly, we use a multi-view volumetric aggregation module for better prediction by exploiting various information in different-scale feature maps. Secondly, in our framework, instead of using all voxels, a mesh-aligned voxel selection module is proposed to make effective prediction by eliminating redundant background voxels. Lastly, the framework further improves the performance of human body parametric modeling by adopting a dual-branch strategy, where one branch for parametric human model prediction and the other for 3D keypoints prediction. Their mutual influence is critical to the improvement for both tasks. Additionally, we find the scarcity of datasets also hinders the development of multi-view methods, so we propose a approach for creating occlusion datasets specifically for multi-view occlusion case. Experimental results verify the effectiveness of the proposed framework on two benchmarks, Human3.6M and MPI-INF-3DHP.

Complete 3D Relationships Extraction Modality Alignment Network for 3D Dense Captioning.

3D dense captioning aims to semantically describe each object detected in a 3D scene, which plays a significant role in 3D scene understanding. Previous works lack a complete definition of 3D spatial relationships and the directly integrate visual and language modalities, thus ignoring the discrepancies between the two modalities. To address these issues, we propose a novel complete 3D relationship extraction modality alignment network, which consists of three steps: 3D object detection, complete 3D relationships extraction, and modality alignment caption. To comprehensively capture the 3D spatial relationship features, we define a complete set of 3D spatial relationships, including the local spatial relationship between objects and the global spatial relationship between each object and the entire scene. To this end, we propose a complete 3D relationships extraction module based on message passing and self-attention to mine multi-scale spatial relationship features and inspect the transformation to obtain features in different views. In addition, we propose the modality alignment caption module to fuse multi-scale relationship features and generate descriptions to bridge the semantic gap from the visual space to the language space with the prior information in the word embedding, and help generate improved descriptions for the 3D scene. Extensive experiments demonstrate that the proposed model outperforms the state-of-the-art methods on the ScanRefer and Nr3D datasets.

MDF-Net: A multi-view dual-attention fusion network for efficient bird sound classification

Bird sound serves as a crucial means of acoustic communication for birds, and its classification research is conducive to the protection, health, and diversity of the ecological ecosystems. Using various feature extraction methods to extract multi-view features can provide more comprehensive information about bird sound, which is a potential method to improve the accuracy of bird sound classification. However, efficiently fusing multi-view features to identify birds accurately remains a challenging task. To address this problem, this paper presents an efficient bird sound classification framework called MDF-Net. The approach extracts four acoustic features from bird sound audios, including wavelet transform spectrogram, Hilbert-Huang transform spectrogram, short-time Fourier transform spectrogram, and Mel-frequency cepstral coefficients, to fully describe the characteristics of bird sound from different views. Subsequently, convolutional neural network is used as advanced feature extractor to obtain deep features of these spectrograms. Then, the multi-head self-attention mechanism focuses on the correlation and importance of different features in each view to obtain essential and expressive feature representations. And the cross-attention mechanism is employed to align and correlate information in the four views, which makes it easier for the classifier to understand the relationships between features of different views. Finally, combined with the results of the dual-attention mechanism, a multi-view fusion feature with difference and diversity is constructed, and it applied to the bird sound classification. In this study, audios from16 bird species constitute the dataset. The multi-view fusion feature based on MDF-Net achieved a classification accuracy of 97.29%, outperformed the 9 single features and 3 fused features used in the experiments. The result demonstrate that the proposed MDF-Net successfully captures the feature relationships within single-view and between multi-view, providing crucial information for correctly classifying bird sound samples. The approach efficiently fuses the features of different views and improves the performance of bird sound classification.

Open-Set Sheep Face Recognition in Multi-View Based on Li-SheepFaceNet

Deep learning-based sheep face recognition improves the efficiency and effectiveness of individual sheep recognition and provides technical support for the development of intelligent livestock farming. However, frequent changes within the flock and variations in facial features in different views significantly affect the practical application of sheep face recognition. In this study, we proposed the Li-SheepFaceNet, a method for open-set sheep face recognition in multi-view. Specifically, we employed the Seesaw block to construct a lightweight model called SheepFaceNet, which significantly improves both performance and efficiency. To enhance the convergence and performance of low-dimensional embedded feature learning, we used Li-ArcFace as the loss function. The Li-SheepFaceNet achieves an open-set recognition accuracy of 96.13% on a self-built dataset containing 3801 multi-view face images of 212 Ujumqin sheep, which surpasses other open-set sheep face recognition methods. To evaluate the robustness and generalization of our approach, we conducted performance testing on a publicly available dataset, achieving a recognition accuracy of 93.33%. Deploying Li-SheepFaceNet on an open-set sheep face recognition system enables the rapid and accurate identification of individual sheep, thereby accelerating the development of intelligent sheep farming.

The real-time dynamic liquid level calculation method of the sucker rod well based on multi-view features fusion

In the production of the sucker rod well, the dynamic liquid level is important for the production efficiency and safety in the lifting process. It is influenced by multi-source data which need to be combined for the dynamic liquid level real-time calculation. In this paper, the multi-source data are regarded as the different views including the load of the sucker rod and liquid in the wellbore, the image of the dynamometer card and production dynamics parameters. These views can be fused by the multi-branch neural network with special fusion layer. With this method, the features of different views can be extracted by considering the difference of the modality and physical meaning between them. Then, the extraction results which are selected by multinomial sampling can be the input of the fusion layer. During the fusion process, the availability under different views determines whether the views are fused in the fusion layer or not. In this way, not only the correlation between the views can be considered, but also the missing data can be processed automatically. The results have shown that the load and production features fusion (the method proposed in this paper) performs best with the lowest mean absolute error (MAE) 39.63 m, followed by the features concatenation with MAE 42.47 m. They both performed better than only a single view and the lower MAE of the features fusion indicates that its generalization ability is stronger. In contrast, the image feature as a single view contributes little to the accuracy improvement after fused with other views with the highest MAE. When there is data missing in some view, compared with the features concatenation, the multi-view features fusion will not result in the unavailability of a large number of samples. When the missing rate is 10%, 30%, 50% and 80%, the method proposed in this paper can reduce MAE by 5.8, 7, 9.3 and 20.3 m respectively. In general, the multi-view features fusion method proposed in this paper can improve the accuracy obviously and process the missing data effectively, which helps provide technical support for real-time monitoring of the dynamic liquid level in oil fields.

Stacked Multiscale Densely Connected Temporal Convolutional Attention Network for Multi-Objective Speech Enhancement in an Airborne Environment

Airborne speech enhancement is always a major challenge for the security of airborne systems. Recently, multi-objective learning technology has become one of the mainstream methods of monaural speech enhancement. In this paper, we propose a novel multi-objective method for airborne speech enhancement, called the stacked multiscale densely connected temporal convolutional attention network (SMDTANet). More specifically, the core of SMDTANet includes three parts, namely a stacked multiscale feature extractor, a triple-attention-based temporal convolutional neural network (TA-TCNN), and a densely connected prediction module. The stacked multiscale feature extractor is leveraged to capture comprehensive feature information from noisy log-power spectra (LPS) inputs. Then, the TA-TCNN adopts a combination of these multiscale features and noisy amplitude modulation spectrogram (AMS) features as inputs to improve its powerful temporal modeling capability. In TA-TCNN, we integrate the advantages of channel attention, spatial attention, and T-F attention to design a novel triple-attention module, which can guide the network to suppress irrelevant information and emphasize informative features of different views. The densely connected prediction module is used to reliably control the flow of the information to provide an accurate estimation of clean LPS and the ideal ratio mask (IRM). Moreover, a new joint-weighted (JW) loss function is constructed to further improve the performance without adding to the model complexity. Extensive experiments on real-world airborne conditions show that our SMDTANet can obtain an on-par or better performance compared to other reference methods in terms of all the objective metrics of speech quality and intelligibility.

Anchor-based multi-view subspace clustering with hierarchical feature descent

Multi-view clustering has attracted growing attention owing to its capabilities of aggregating information from various sources and its promising horizons in public affairs. Up till now, many advanced approaches have been proposed in recent literature. However, there are several ongoing difficulties to be tackled. One common dilemma occurs while attempting to align the features of different views. Moreover, due to the fact that many existing multi-view clustering algorithms stem from spectral clustering, this results to cubic time complexity w.r.t. the number of dataset. However, we propose Anchor-based Multi-view Subspace Clustering with Hierarchical Feature Descent(MVSC-HFD) to tackle the discrepancy among views through hierarchical feature descent and project to a common subspace( STAGE 1), which reveals dependency of different views. We further reduce the computational complexity to linear time cost through a unified sampling strategy in the common subspace( STAGE 2), followed by anchor-based subspace clustering to learn the bipartite graph collectively( STAGE 3). Extensive experimental results on public benchmark datasets demonstrate that our proposed model consistently outperforms the state-of-the-art techniques. Our code is publicly available at https://github.com/QiyuanOu/MVSC-HFD/tree/main.

Sequential visual and semantic consistency for semi-supervised text recognition

Scene text recognition (STR) is a challenging task that requires large-scale annotated data for training. However, collecting and labeling real text images is expensive and time-consuming, which limits the availability of real data. Therefore, most existing STR methods resort to synthetic data, which may introduce domain discrepancy and degrade the performance of STR models. To alleviate this problem, recent semi-supervised STR methods exploit unlabeled real data by enforcing character-level consistency regularization between weakly and strongly augmented views of the same image. However, these methods neglect word-level consistency, which is crucial for sequence recognition tasks. This paper proposes a novel semi-supervised learning method for STR that incorporates word-level consistency regularization from both visual and semantic aspects. Specifically, we devise a shortest path alignment module to align the sequential visual features of different views and minimize their distance. Moreover, we adopt a reinforcement learning framework to optimize the semantic similarity of the predicted strings in the embedding space. We conduct extensive experiments on several standard and challenging STR benchmarks and demonstrate the superiority of our proposed method over existing semi-supervised STR methods.

Partial Multiview Representation Learning With Cross-View Generation.

Multiview learning has made significant progress in recent years. However, an implicit assumption is that multiview data are complete, which is often contrary to practical applications. Due to human or data acquisition equipment errors, what we actually get is partial multiview data, which existing multiview algorithms are limited to processing. Modeling complex dependencies between views in terms of consistency and complementarity remains challenging, especially in partial multiview data scenarios. To address the above issues, this article proposes a deep Gaussian cross-view generation model (named PMvCG), which aims to model views according to the principles of consistency and complementarity and eventually learn the comprehensive representation of partial multiview data. PMvCG can discover cross-view associations by learning view-sharing and view-specific features of different views in the representation space. The missing views can be reconstructed and are applied in turn to further optimize the model. The estimated uncertainty in the model is also considered and integrated into the representation to improve the performance. We design a variational inference and iterative optimization algorithm to solve PMvCG effectively. We conduct comprehensive experiments on multiple real-world datasets to validate the performance of PMvCG. We compare the PMvCG with various methods by applying the learned representation to clustering and classification. We also provide more insightful analysis to explore the PMvCG, such as convergence analysis, parameter sensitivity analysis, and the effect of uncertainty in the representation. The experimental results indicate that PMvCG obtains promising results and surpasses other comparative methods under different experimental settings.

Multi-view convolutional vision transformer for 3D object recognition

With the rapid development of three-dimensional (3D) vision technology and the increasing application of 3D objects, there is an urgent need for 3D object recognition in the fields of computer vision, virtual reality, and artificial intelligence robots. The view-based method projects 3D objects into two-dimensional (2D) images from different viewpoints and applies convolutional neural networks (CNN) to model the projected views. Although these methods have achieved excellent recognition performance, there is not sufficient information interaction between the features of different views in these methods. Inspired by the recent success achieved by vision transformer (ViT) in image recognition, we propose a hybrid network by taking advantage of CNN to extract multi-scale local information of each view, and of transformer to capture the relevance of multi-scale information between different views. To verify the effectiveness of our multi-view convolutional vision transformer (MVCVT), we conduct experiments on two public benchmarks, ModelNet40 and ModelNet10, and compare with those of some state-of-the-art methods. The final results show that MVCVT has competitive performance in 3D object recognition.

3D multi-view squeeze-and-excitation convolutional neural network for lung nodule classification.

Early screening is crucial to improve the survival rate and recovery rate of lung cancer patients. Computer-aided diagnosis system (CAD) is a powerful tool to assist clinicians in early diagnosis. Lung nodules are characterized by spatial heterogeneity. However, many attempts use the two-dimensional multi-view (MV) framework to learn and simply integrate multiple view features. These methods suffer from the problems of not capturing the spatial characteristics effectively and ignoring the variability of multiple views. In this paper, we propose a three-dimensional MV convolutional neural network (3D MVCNN) framework and embed the squeeze-and-excitation (SE) module in it to further address the variability of each view in the MV framework. First, the 3D multiple view samples of lung nodules are extracted by the spatial sampling method, and a 3D CNN is established to extract 3D abstract features. Second, build a 3D MVCNN framework according to the 3D multiple view samples and 3D CNN. This framework can learn more features of different views of lung nodules, taking into account the characteristics of spatial heterogeneity of lung nodules. Finally, to further address the variability of each view in the MV framework, a 3D MVSECNN model is constructed by introducing a SE module in the feature fusion stage. For training and testing purposes we used independent subsets of the public LIDC-IDRI dataset. For the LIDC-IDRI dataset, this study achieved 96.04% accuracy and 98.59% sensitivity in the binary classification, and 87.76% accuracy in the ternary classification, which was higher than other state-of-the-art studies. The consistency score of 0.948 between the model predictions and pathological diagnosis was significantly higher than that between the clinician's annotations and pathological diagnosis. The results show that our proposed method can effectively learn the spatial heterogeneity of nodules and solve the problem of multiple view variability. Moreover, the consistency analysis indicates that our method can provide clinicians with more accurate results of benign-malignant lung nodule classification for auxiliary diagnosis, which is important for assisting clinicians in clinical diagnosis.

Stereo Refinement Dehazing Network

The performance of stereo vision tasks degrades when haze exists in the input stereo image pair. Independently applying single image dehazing algorithm on left and right images is not optimal. To overcome the problem, we propose an effective framework, called SRDNet, for simultaneously dehazing stereo images. The main idea of SRDNet is to make full use of the stereo information from cross views improving dehazing performance. It does not explicitly employ the disparity estimation and the correlation matrix. SRDNet comprises two parts: a weight-sharing coarse dehazing network (WSCDN) and a guided separated refinement network (GSRN). The WSCDN is utilized to predict a coarse dehazed image pair. Then the GSRN is introduced to predict the residues for different views by extracting the fused information of cross views and separating the features of different views with a guided channel and spatial refinement module. The residues are added to the coarse dehazed pair so as to make refinement and remove the remained haze. Experimental results demonstrate that our proposed SRDNet surpasses previous image dehazing methods by a significant margin both quantitatively and qualitatively. Moreover, our SRDNet could be a preprocessing step of the stereo-based 3D object detection and boosts the 3D detection accuracy in hazy scenes.

Multiorder Interaction Information Embedding-Based Multiview Fusion-Aided Hyperspectral Image Classification

Hyperspectral images (HSI) are obtained from hyperspectral imaging sensors, which capture information in hundreds of spectral bands of objects. However, how to take full advantage of spatial and spectral information from many spectral bands to improve the performance of HSI classification remains an open question. Many HSI classification works have recently been reported by employing multi-view learning (MVL) algorithms that can fully use complementary information between different view features and thus have received widespread attention. This paper proposes a multi-view fusion network based on multi-order interaction information embedding for HSI classification. Firstly, the correlation matrix between spectral bands is used to divide the original data into multiple subsets as local views. The subset after the Segmented-PCA process is used as the global view. Secondly, the features of different views are extracted separately using a feature extraction network and mapped to the same dimension. Pre-fusion is achieved by multi-order interaction of various view features. Finally, loss-weighted fusion is applied to each view according to its contribution to the classification task. To evaluate the effectiveness of the proposed method, complete experiments were conducted on three commonly used HSI datasets, namely Pavia University, Houston 2013, and Houston 2018. The experimental results demonstrate that the proposed method improves the classification performance of existing feature extraction networks and is more competitive with other methods in the field.

A Faster and More Effective Cross-View Matching Method of UAV and Satellite Images for UAV Geolocalization

Cross-view geolocalization matches the same target in different images from various views, such as views of unmanned aerial vehicles (UAVs) and satellites, which is a key technology for UAVs to autonomously locate and navigate without a positioning system (e.g., GPS and GNSS). The most challenging aspect in this area is the shifting of targets and nonuniform scales among different views. Published methods focus on extracting coarse features from parts of images, but neglect the relationship between different views, and the influence of scale and shifting. To bridge this gap, an effective network is proposed with well-designed structures, referred to as multiscale block attention (MSBA), based on a local pattern network. MSBA cuts images into several parts with different scales, among which self-attention is applied to make feature extraction more efficient. The features of different views are extracted by a multibranch structure, which was designed to make different branches learn from each other, leading to a more subtle relationship between views. The method was implemented with the newest UAV-based geolocalization dataset. Compared with the existing state-of-the-art (SOTA) method, MSBA accuracy improved by almost 10% when the inference time was equal to that of the SOTA method; when the accuracy of MSBA was the same as that of the SOTA method, inference time was shortened by 30%.

Multilevel projections with adaptive neighbor graph for unsupervised multi-view feature selection

Multi-view feature selection aims at obtaining a subset of informative features from heterogeneous feature domains. Recent graph based approaches mostly learn view-specific feature selection matrices by virtue of prepared single-view graphs, and weight the view-wise objectives to discriminate them. However, the majority of them encounter that (i) the dimensions of vectors for evaluating features in different views are inconsistent and (ii) the similarities attained by using features in the original high-dimensional space are inferior. As a result, the joint evaluation of heterogeneous features using view-specific selection matrices are inaccurate and immensely depend on ill-defined similarity relations. To overcome them, we propose the Multilevel projections with Adaptive neighbor graph model for unsupervised Multi-view Feature Selection (MAMFS). Our formulation learns the collaborative graph adapting to the adaptive k-nearest neighbors in subspaces, wherein the multilevel projections are involved including the weighted view-specific projections that discriminate different views and the joint projection that collaborates different views. Benefited from this, both of view-within information and the view-across information are jointly explored and made use of. Extensive simulations are conducted to validate the superiority of the proposed method compared to the state-of-the-art competitors.

Auto-Weighted Multi-View Discriminative Metric Learning Method With Fisher Discriminative and Global Structure Constraints for Epilepsy EEG Signal Classification.

Metric learning is a class of efficient algorithms for EEG signal classification problem. Usually, metric learning method deals with EEG signals in the single view space. To exploit the diversity and complementariness of different feature representations, a new auto-weighted multi-view discriminative metric learning method with Fisher discriminative and global structure constraints for epilepsy EEG signal classification called AMDML is proposed to promote the performance of EEG signal classification. On the one hand, AMDML exploits the multiple features of different views in the scheme of the multi-view feature representation. On the other hand, considering both the Fisher discriminative constraint and global structure constraint, AMDML learns the discriminative metric space, in which the intraclass EEG signals are compact and the interclass EEG signals are separable as much as possible. For better adjusting the weights of constraints and views, instead of manually adjusting, a closed form solution is proposed, which obtain the best values when achieving the optimal model. Experimental results on Bonn EEG dataset show AMDML achieves the satisfactory results.

Multi-View Features Joint Learning with Label and Local Distribution Consistency for Point Cloud Classification

In outdoor Light Detection and Ranging (lidar)point cloud classification, finding the discriminative features for point cloud perception and scene understanding represents one of the great challenges. The features derived from defect-laden (i.e., noise, outliers, occlusions and irregularities) and raw outdoor LiDAR scans usually contain redundant and irrelevant information which adversely affects the accuracy of point semantic labeling. Moreover, point cloud features of different views have a capability to express different attributes of the same point. The simplest way of concatenating these features of different views cannot guarantee the applicability and effectiveness of the fused features. To solve these problems and achieve outdoor point cloud classification with fewer training samples, we propose a novel multi-view features and classifiers’ joint learning framework. The proposed framework uses label consistency and local distribution consistency of multi-space constraints for multi-view point cloud features extraction and classification. In the framework, the manifold learning is used to carry out subspace joint learning of multi-view features by introducing three kinds of constraints, i.e., local distribution consistency of feature space and position space, label consistency among multi-view predicted labels and ground truth, and label consistency among multi-view predicted labels. The proposed model can be well trained by fewer training points, and an iterative algorithm is used to solve the joint optimization of multi-view feature projection matrices and linear classifiers. Subsequently, the multi-view features are fused and used for point cloud classification effectively. We evaluate the proposed method on five different point cloud scenes and experimental results demonstrate that the classification performance of the proposed method is at par or outperforms the compared algorithms.

Depth prediction from 2D images: A taxonomy and an evaluation study

Among the various cues that help us understand and interact with our surroundings, depth is of particular importance. It allows us to move in space and grab objects to complete different tasks. Therefore, depth prediction has been an active research field for decades and many algorithms have been proposed to retrieve depth. Some imitate human vision and compute depth through triangulation on correspondences found between pixels or handcrafted features in different views of the same scene. Others rely on simple assumptions and semantic knowledge of the structure of the scene to get the depth information. Recently, numerous algorithms based on deep learning have emerged from the computer vision community. They implement the same principles as the non-deep learning methods and leverage the ability of deep neural networks of automatically learning important features that help to solve the task. By doing so, they produce new state-of-the-art results and show encouraging prospects. In this article, we propose a taxonomy of deep learning methods for depth prediction from 2D images. We retained the training strategy as the sorting criterion. Indeed, some methods are trained in a supervised manner which means depth labels are needed during training while others are trained in an unsupervised manner. In that case, the models learn to perform a different task such as view synthesis and depth is only a by-product of this learning. In addition to this taxonomy, we also evaluate nine models on two similar datasets without retraining. Our analysis showed that (i) most models are sensitive to sharp discontinuities created by shadows or colour contrasts and (ii) the post processing applied to the results before computing the commonly used metrics can change the model ranking. Moreover, we showed that most metrics agree with each other and are thus redundant.