Joni-Kristian Kämäräinen

Lauri Suomela, Jussi Kalliola, Atakan Dag et al.

This work introduces a simulator-based benchmark for visual localization in the autonomous navigation context. The dynamic benchmark enables investigation of how variables such as the time of day, weather, and camera perspective affect the navigation performance of autonomous agents that utilize visual localization for closed-loop control. The experimental part of the paper studies the effects of four such variables by evaluating state-of-the-art visual localization methods as part of the motion planning module of an autonomous navigation stack. The results show major variation in the suitability of the different methods for vision-based navigation. To the authors' best knowledge, the proposed benchmark is the first to study modern visual localization methods as part of a complete navigation stack. We make the benchmark available at https://github.com/lasuomela/carla_vloc_benchmark.

Jinyu Yang, Zhe Li, Song Yan et al.

RGBD object tracking is gaining momentum in computer vision research thanks to the development of depth sensors. Although numerous RGBD trackers have been proposed with promising performance, an in-depth review for comprehensive understanding of this area is lacking. In this paper, we firstly review RGBD object trackers from different perspectives, including RGBD fusion, depth usage, and tracking framework. Then, we summarize the existing datasets and the evaluation metrics. We benchmark a representative set of RGBD trackers, and give detailed analyses based on their performances. Particularly, we are the first to provide depth quality evaluation and analysis of tracking results in depth-friendly scenarios in RGBD tracking. For long-term settings in most RGBD tracking videos, we give an analysis of trackers' performance on handling target disappearance. To enable better understanding of RGBD trackers, we propose robustness evaluation against input perturbations. Finally, we summarize the challenges and provide open directions for this community. All resources are publicly available at https://github.com/memoryunreal/RGBD-tracking-review.

Lauri Suomela, Jussi Kalliola, Harry Edelman et al.

Recent results suggest that splitting topological navigation into robot-independent and robot-specific components improves navigation performance by enabling the robot-independent part to be trained with data collected by robots of different types. However, the navigation methods' performance is still limited by the scarcity of suitable training data and they suffer from poor computational scaling. In this work, we present PlaceNav, subdividing the robot-independent part into navigation-specific and generic computer vision components. We utilize visual place recognition for the subgoal selection of the topological navigation pipeline. This makes subgoal selection more efficient and enables leveraging large-scale datasets from non-robotics sources, increasing training data availability. Bayesian filtering, enabled by place recognition, further improves navigation performance by increasing the temporal consistency of subgoals. Our experimental results verify the design and the new method obtains a 76% higher success rate in indoor and 23% higher in outdoor navigation tasks with higher computational efficiency.

German F. Torres, Joni-Kristian Kämäräinen

Camera motion introduces spatially varying blur due to the depth changes in the 3D world. This work investigates scene configurations where such blur is produced under parallax camera motion. We present a simple, yet accurate, Image Compositing Blur (ICB) model for depth-dependent spatially varying blur. The (forward) model produces realistic motion blur from a single image, depth map, and camera trajectory. Furthermore, we utilize the ICB model, combined with a coordinate-based MLP, to learn a sharp neural representation from the blurred input. Experimental results are reported for synthetic and real examples. The results verify that the ICB forward model is computationally efficient and produces realistic blur, despite the lack of occlusion information. Additionally, our method for restoring a sharp representation proves to be a competitive approach for the deblurring task.

Song Yan, Jinyu Yang, Jani Käpylä et al.

RGBD (RGB plus depth) object tracking is gaining momentum as RGBD sensors have become popular in many application fields such as robotics.However, the best RGBD trackers are extensions of the state-of-the-art deep RGB trackers. They are trained with RGB data and the depth channel is used as a sidekick for subtleties such as occlusion detection. This can be explained by the fact that there are no sufficiently large RGBD datasets to 1) train deep depth trackers and to 2) challenge RGB trackers with sequences for which the depth cue is essential. This work introduces a new RGBD tracking dataset - Depth-Track - that has twice as many sequences (200) and scene types (40) than in the largest existing dataset, and three times more objects (90). In addition, the average length of the sequences (1473), the number of deformable objects (16) and the number of annotated tracking attributes (15) have been increased. Furthermore, by running the SotA RGB and RGBD trackers on DepthTrack, we propose a new RGBD tracking baseline, namely DeT, which reveals that deep RGBD tracking indeed benefits from genuine training data. The code and dataset is available at https://github.com/xiaozai/DeT

German F. Torres, Jussi Kalliola, Soumya Tripathy et al.

Video deblurring aims at recovering sharp details from a sequence of blurry frames. Despite the proliferation of depth sensors in mobile phones and the potential of depth information to guide deblurring, depth-aware deblurring has received only limited attention. In this work, we introduce the 'Depth-Aware VIdeo DEblurring' (DAVIDE) dataset to study the impact of depth information in video deblurring. The dataset comprises synchronized blurred, sharp, and depth videos. We investigate how the depth information should be injected into the existing deep RGB video deblurring models, and propose a strong baseline for depth-aware video deblurring. Our findings reveal the significance of depth information in video deblurring and provide insights into the use cases where depth cues are beneficial. In addition, our results demonstrate that while the depth improves deblurring performance, this effect diminishes when models are provided with a longer temporal context. Project page: https://germanftv.github.io/DAVIDE.github.io/ .

Joni-Kristian Kämäräinen

Understanding the transformer architecture and its workings is essential for machine learning (ML) engineers. However, truly understanding the transformer architecture can be demanding, even if you have a solid background in machine learning or deep learning. The main working horse is attention, which yields to the transformer encoder-decoder structure. However, putting attention aside leaves several programming components that are easy to implement but whose role for the whole is unclear. These components are 'tokenization', 'embedding' ('un-embedding'), 'masking', 'positional encoding', and 'padding'. The focus of this work is on understanding them. To keep things simple, the understanding is built incrementally by adding components one by one, and after each step investigating what is doable and what is undoable with the current model. Simple sequences of zeros (0) and ones (1) are used to study the workings of each step.

Lauri Suomela, Sasanka Kuruppu Arachchige, German F. Torres et al.

This paper investigates how the performance of visual navigation policies trained in simulation compares to policies trained with real-world data. Performance degradation of simulator-trained policies is often significant when they are evaluated in the real world. However, despite this well-known sim-to-real gap, we demonstrate that simulator-trained policies can match the performance of their real-world-trained counterparts. Central to our approach is a navigation policy architecture that bridges the sim-to-real appearance gap by leveraging pretrained visual representations and runs real-time on robot hardware. Evaluations on a wheeled mobile robot show that the proposed policy, when trained in simulation, outperforms its real-world-trained version by 31% and the prior state-of-the-art methods by 50% in navigation success rate. Policy generalization is verified by deploying the same model onboard a drone. Our results highlight the importance of diverse image encoder pretraining for sim-to-real generalization, and identify on-policy learning as a key advantage of simulated training over training with real data.

Joni-Kristian Kämäräinen

Transformer is the state-of-the-art model for many natural language processing, computer vision, and audio analysis problems. Transformer effectively combines information from the past input and output samples in auto-regressive manner so that each sample becomes aware of all inputs and outputs. In sequence-to-sequence (Seq2Seq) modeling, the transformer processed samples become effective in predicting the next output. Time series forecasting is a Seq2Seq problem. The original architecture is defined for discrete input and output sequence tokens, but to adopt it for time series, the model must be adapted for continuous data. This work introduces minimal adaptations to make the original transformer architecture suitable for continuous value time series data.

Vivienne Huiling Wang, Tinghuai Wang, Wenyan Yang et al.

In goal-conditioned hierarchical reinforcement learning (HRL), a high-level policy specifies a subgoal for the low-level policy to reach. Effective HRL hinges on a suitable subgoal represen tation function, abstracting state space into latent subgoal space and inducing varied low-level behaviors. Existing methods adopt a subgoal representation that provides a deterministic mapping from state space to latent subgoal space. Instead, this paper utilizes Gaussian Processes (GPs) for the first probabilistic subgoal representation. Our method employs a GP prior on the latent subgoal space to learn a posterior distribution over the subgoal representation functions while exploiting the long-range correlation in the state space through learnable kernels. This enables an adaptive memory that integrates long-range subgoal information from prior planning steps allowing to cope with stochastic uncertainties. Furthermore, we propose a novel learning objective to facilitate the simultaneous learning of probabilistic subgoal representations and policies within a unified framework. In experiments, our approach outperforms state-of-the-art baselines in standard benchmarks but also in environments with stochastic elements and under diverse reward conditions. Additionally, our model shows promising capabilities in transferring low-level policies across different tasks.

Vivienne Huiling Wang, Joni Pajarinen, Tinghuai Wang et al.

Hierarchical reinforcement learning (HRL) proposes to solve difficult tasks by performing decision-making and control at successively higher levels of temporal abstraction. However, off-policy HRL often suffers from the problem of a non-stationary high-level policy since the low-level policy is constantly changing. In this paper, we propose a novel HRL approach for mitigating the non-stationarity by adversarially enforcing the high-level policy to generate subgoals compatible with the current instantiation of the low-level policy. In practice, the adversarial learning is implemented by training a simple state-conditioned discriminator network concurrently with the high-level policy which determines the compatibility level of subgoals. Comparison to state-of-the-art algorithms shows that our approach improves both learning efficiency and performance in challenging continuous control tasks.

Wenyan Yang, Nataliya Strokina, Nikolay Serbenyuk et al.

We have recently proposed two pile loading controllers that learn from human demonstrations: a neural network (NNet) [1] and a random forest (RF) controller [2]. In the field experiments the RF controller obtained clearly better success rates. In this work, the previous findings are drastically revised by experimenting summer time trained controllers in winter conditions. The winter experiments revealed a need for additional sensors, more training data, and a controller that can take advantage of these. Therefore, we propose a revised neural controller (NNetV2) which has a more expressive structure and uses a neural attention mechanism to focus on important parts of the sensor and control signals. Using the same data and sensors to train and test the three controllers, NNetV2 achieves better robustness against drastically changing conditions and superior success rate. To the best of our knowledge, this is the first work testing a learning-based controller for a heavy-duty machine in drastically varying outdoor conditions and delivering high success rate in winter, being trained in summer.

Song Yan, Joni-Kristian Kämäräinen

Accurate estimation of anthropometric body measurements from RGB images has many potential applications in industrial design, online clothing, medical diagnosis and ergonomics. Research on this topic is limited by the fact that there exist only generated datasets which are based on fitting a 3D body mesh to 3D body scans in the commercial CAESAR dataset. For 2D only silhouettes are generated. To circumvent the data bottleneck, we introduce a new 3D scan dataset of 2,675 female and 1,474 male scans. We also introduce a small dataset of 200 RGB images and tape measured ground truth. With the help of the two new datasets we propose a part-based shape model and a deep neural network for estimating anthropometric measurements from 2D images. All data will be made publicly available.

Yanlin Qian, Miaojing Shi, Joni-Kristian Kämäräinen et al.

We address the problem of decomposing an image into albedo and shading. We propose the Fast Fourier Intrinsic Network, FFI-Net in short, that operates in the spectral domain, splitting the input into several spectral bands. Weights in FFI-Net are optimized in the spectral domain, allowing faster convergence to a lower error. FFI-Net is lightweight and does not need auxiliary networks for training. The network is trained end-to-end with a novel spectral loss which measures the global distance between the network prediction and corresponding ground truth. FFI-Net achieves state-of-the-art performance on MPI-Sintel, MIT Intrinsic, and IIW datasets.

Yanlin Qian, Jani Käpylä, Joni-Kristian Kämäräinen et al.

Temporal Color Constancy (CC) is a recently proposed approach that challenges the conventional single-frame color constancy. The conventional approach is to use a single frame - shot frame - to estimate the scene illumination color. In temporal CC, multiple frames from the view finder sequence are used to estimate the color. However, there are no realistic large scale temporal color constancy datasets for method evaluation. In this work, a new temporal CC benchmark is introduced. The benchmark comprises of (1) 600 real-world sequences recorded with a high-resolution mobile phone camera, (2) a fixed train-test split which ensures consistent evaluation, and (3) a baseline method which achieves high accuracy in the new benchmark and the dataset used in previous works. Results for more than 20 well-known color constancy methods including the recent state-of-the-arts are reported in our experiments.

Yanlin Qian, Alan Lukežič, Matej Kristan et al.

The best RGBD trackers provide high accuracy but are slow to run. On the other hand, the best RGB trackers are fast but clearly inferior on the RGBD datasets. In this work, we propose a deep depth-aware long-term tracker that achieves state-of-the-art RGBD tracking performance and is fast to run. We reformulate deep discriminative correlation filter (DCF) to embed the depth information into deep features. Moreover, the same depth-aware correlation filter is used for target re-detection. Comprehensive evaluations show that the proposed tracker achieves state-of-the-art performance on the Princeton RGBD, STC, and the newly-released CDTB benchmarks and runs 20 fps.

Song Yan, Johan Wirta, Joni-Kristian Kämäräinen

We propose a full processing pipeline to acquire anthropometric measurements from 3D measurements. The first stage of our pipeline is a commercial point cloud scanner. In the second stage, a pre-defined body model is fitted to the captured point cloud. We have generated one male and one female model from the SMPL library. The fitting process is based on non-rigid Iterative Closest Point (ICP) algorithm that minimizes overall energy of point distance and local stiffness energy terms. In the third stage, we measure multiple circumference paths on the fitted model surface and use a non-linear regressor to provide the final estimates of anthropometric measurements. We scanned 194 male and 181 female subjects and the proposed pipeline provides mean absolute errors from 2.5 mm to 16.0 mm depending on the anthropometric measurement.

Antti Hietanen, Jyrki Latokartano, Roel Pieters et al.

Industrial standards define safety requirements for Human-Robot Collaboration (HRC) in industrial manufacturing. The standards particularly require real-time monitoring and securing of the minimum protective distance between a robot and an operator. In this work, we propose a depth-sensor based model for workspace monitoring and an interactive Augmented Reality (AR) User Interface (UI) for safe HRC. The AR UI is implemented on two different hardware: a projector-mirror setup anda wearable AR gear (HoloLens). We experiment the workspace model and UIs for a realistic diesel motor assembly task. The AR-based interactive UIs provide 21-24% and 57-64% reduction in the task completion and robot idle time, respectively, as compared to a baseline without interaction and workspace sharing. However, subjective evaluations reveal that HoloLens based AR is not yet suitable for industrial manufacturing while the projector-mirror setup shows clear improvements in safety and work ergonomics.

Alan Lukežič, Ugur Kart, Jani Käpylä et al.

A long-term visual object tracking performance evaluation methodology and a benchmark are proposed. Performance measures are designed by following a long-term tracking definition to maximize the analysis probing strength. The new measures outperform existing ones in interpretation potential and in better distinguishing between different tracking behaviors. We show that these measures generalize the short-term performance measures, thus linking the two tracking problems. Furthermore, the new measures are highly robust to temporal annotation sparsity and allow annotation of sequences hundreds of times longer than in the current datasets without increasing manual annotation labor. A new challenging dataset of carefully selected sequences with many target disappearances is proposed. A new tracking taxonomy is proposed to position trackers on the short-term/long-term spectrum. The benchmark contains an extensive evaluation of the largest number of long-term tackers and comparison to state-of-the-art short-term trackers. We analyze the influence of tracking architecture implementations to long-term performance and explore various re-detection strategies as well as influence of visual model update strategies to long-term tracking drift. The methodology is integrated in the VOT toolkit to automate experimental analysis and benchmarking and to facilitate future development of long-term trackers.

Antti Hietanen, Jyrki Latokartano, Alessandro Foi et al.

Vision based object grasping and manipulation in robotics require accurate estimation of object's 6D pose. The 6D pose estimation has received significant attention in computer vision community and multiple datasets and evaluation metrics have been proposed. However, the existing metrics measure how well two geometrical surfaces are aligned - ground truth vs. estimated pose - which does not directly measure how well a robot can perform the task with the given estimate. In this work we propose a probabilistic metric that directly measures success in robotic tasks. The evaluation metric is based on non-parametric probability density that is estimated from samples of a real physical setup. During the pose evaluation stage the physical setup is not needed. The evaluation metric is validated in controlled experiments and a new pose estimation dataset of industrial parts is introduced. The experimental results with the parts confirm that the proposed evaluation metric better reflects the true performance in robotics than the existing metrics.

Yanlin Qian, Song Yan, Joni-Kristian Kämäräinen et al.

In the real world, a scene is usually cast by multiple illuminants and herein we address the problem of spatial illumination estimation. Our solution is based on detecting gray pixels with the help of flash photography. We show that flash photography significantly improves the performance of gray pixel detection without illuminant prior, training data or calibration of the flash. We also introduce a novel flash photography dataset generated from the MIT intrinsic dataset.

Yanlin Qian, Joni-Kristian Kämäräinen, Jarno Nikkanen et al.

We propose a novel grayness index for finding gray pixels and demonstrate its effectiveness and efficiency in illumination estimation. The grayness index, GI in short, is derived using the Dichromatic Reflection Model and is learning-free. GI allows to estimate one or multiple illumination sources in color-biased images. On standard single-illumination and multiple-illumination estimation benchmarks, GI outperforms state-of-the-art statistical methods and many recent deep methods. GI is simple and fast, written in a few dozen lines of code, processing a 1080p image in ~0.4 seconds with a non-optimized Matlab code.

Junsheng Fu, Said Pertuz, Jiri Matas et al.

We consider a single-query 6-DoF camera pose estimation with reference images and a point cloud, i.e. the problem of estimating the position and orientation of a camera by using reference images and a point cloud. In this work, we perform a systematic comparison of three state-of-the-art strategies for 6-DoF camera pose estimation, i.e. feature-based, photometric-based and mutual-information-based approaches. The performance of the studied methods is evaluated on two standard datasets in terms of success rate, translation error and max orientation error. Building on the results analysis, we propose a hybrid approach that combines feature-based and mutual-information-based pose estimation methods since it provides complementary properties for pose estimation. Experiments show that (1) in cases with large environmental variance, the hybrid approach outperforms feature-based and mutual-information-based approaches by an average of 25.1% and 5.8% in terms of success rate, respectively; (2) in cases where query and reference images are captured at similar imaging conditions, the hybrid approach performs similarly as the feature-based approach, but outperforms both photometric-based and mutual-information-based approaches with a clear margin; (3) the feature-based approach is consistently more accurate than mutual-information-based and photometric-based approaches when at least 4 consistent matching points are found between the query and reference images.

Yanlin Qian, Said Pertuz, Jarno Nikkanen et al.

We present a statistical color constancy method that relies on novel gray pixel detection and mean shift clustering. The method, called Mean Shifted Grey Pixel -- MSGP, is based on the observation: true-gray pixels are aligned towards one single direction. Our solution is compact, easy to compute and requires no training. Experiments on two real-world benchmarks show that the proposed approach outperforms state-of-the-art methods in the camera-agnostic scenario. In the setting where the camera is known, MSGP outperforms all statistical methods.

Uğur Kart, Joni-Kristian Kämäräinen, Jiří Matas et al.

Depth information provides a strong cue for occlusion detection and handling, but has been largely omitted in generic object tracking until recently due to lack of suitable benchmark datasets and applications. In this work, we propose a Depth Masked Discriminative Correlation Filter (DM-DCF) which adopts novel depth segmentation based occlusion detection that stops correlation filter updating and depth masking which adaptively adjusts the spatial support for correlation filter. In Princeton RGBD Tracking Benchmark, our DM-DCF is among the state-of-the-art in overall ranking and the winner on multiple categories. Moreover, since it is based on DCF, ``DM-DCF`` runs an order of magnitude faster than its competitors making it suitable for time constrained applications.

Dingding Cai, Ke Chen, Yanlin Qian et al.

Successful fine-grained image classification methods learn subtle details between visually similar (sub-)classes, but the problem becomes significantly more challenging if the details are missing due to low resolution. Encouraged by the recent success of Convolutional Neural Network (CNN) architectures in image classification, we propose a novel resolution-aware deep model which combines convolutional image super-resolution and convolutional fine-grained classification into a single model in an end-to-end manner. Extensive experiments on the Stanford Cars and Caltech-UCSD Birds 200-2011 benchmarks demonstrate that the proposed model consistently performs better than conventional convolutional net on classifying fine-grained object classes in low-resolution images.