This work presents a framework for automatically extracting physical object properties, such as material composition, mass, volume, and stiffness, through robot manipulation and a database of object measurements. The framework involves exploratory action selection to maximize learning about objects on a table. A Bayesian network models conditional dependencies between object properties, incorporating prior probability distributions and uncertainty associated with measurement actions. The algorithm selects optimal exploratory actions based on expected information gain and updates object properties through Bayesian inference. Experimental evaluation demonstrates effective action selection compared to a baseline and correct termination of the experiments if there is nothing more to be learned. The algorithm proved to behave intelligently when presented with trick objects with material properties in conflict with their appearance. The robot pipeline integrates with a logging module and an online database of objects, containing over 24,000 measurements of 63 objects with different grippers. All code and data are publicly available, facilitating automatic digitization of objects and their physical properties through exploratory manipulations.
Monocular depth estimation (MDE) is inherently ambiguous, as a given image may result from many different 3D scenes and vice versa. To resolve this ambiguity, an MDE system must make assumptions about the most likely 3D scenes for a given input. These assumptions can be either explicit or implicit. In this work, we demonstrate the use of natural language as a source of an explicit prior about the structure of the world. The assumption is made that human language encodes the likely distribution in depth-space of various objects. We first show that a language model encodes this implicit bias during training, and that it can be extracted using a very simple learned approach. We then show that this prediction can be provided as an explicit source of assumption to an MDE system, using an off-the-shelf instance segmentation model that provides the labels used as the input to the language model. We demonstrate the performance of our method on the NYUD2 dataset, showing improvement compared to the baseline and to random controls.
Traditional knowledge distillation (KD) relies on a proficient teacher trained on the target task, which is not always available. In this setting, cross-task distillation can be used, enabling the use of any teacher model trained on a different task. However, many KD methods prove ineffective when applied to this cross-task setting. To address this limitation, we propose a simple modification: the use of an inverted projection. We show that this drop-in replacement for a standard projector is effective by learning to disregard any task-specific features which might degrade the student's performance. We find that this simple modification is sufficient for extending many KD methods to the cross-task setting, where the teacher and student tasks can be very different. In doing so, we obtain up to a 1.9% improvement in the cross-task setting compared to the traditional projection, at no additional cost. Our method can obtain significant performance improvements (up to 7%) when using even a randomly-initialised teacher on various tasks such as depth estimation, image translation, and semantic segmentation, despite the lack of any learned knowledge to transfer. To provide conceptual and analytical insights into this result, we show that using an inverted projection allows the distillation loss to be decomposed into a knowledge transfer and a spectral regularisation component. Through this analysis we are additionally able to propose a novel regularisation loss that allows teacher-free distillation, enabling performance improvements of up to 8.57% on ImageNet with no additional training costs.
Acquiring and utilizing accurate channel state information (CSI) can significantly improve transmission performance, thereby holding a crucial role in realizing the potential advantages of massive multiple-input multiple-output (MIMO) technology. Current prevailing CSI feedback approaches improve precision by employing advanced deep-learning methods to learn representative CSI features for a subsequent compression process. Diverging from previous works, we treat the CSI compression problem in the context of implicit neural representations. Specifically, each CSI matrix is viewed as a neural function that maps the CSI coordinates (antenna number and subchannel) to the corresponding channel gains. Instead of transmitting the parameters of the implicit neural functions directly, we transmit modulations based on the CSI matrix derived through a meta-learning algorithm. Modulations are then applied to a shared base network to generate the elements of the CSI matrix. Modulations corresponding to the CSI matrix are quantized and entropy-coded to further reduce the communication bandwidth, thus achieving extreme CSI compression ratios. Numerical results show that our proposed approach achieves state-of-the-art performance and showcases flexibility in feedback strategies.
Dynamic stereo matching is the task of estimating consistent disparities from stereo videos with dynamic objects. Recent learning-based methods prioritize optimal performance on a single stereo pair, resulting in temporal inconsistencies. Existing video methods apply per-frame matching and window-based cost aggregation across the time dimension, leading to low-frequency oscillations at the scale of the window size. Towards this challenge, we develop a bidirectional alignment mechanism for adjacent frames as a fundamental operation. We further propose a novel framework, BiDAStereo, that achieves consistent dynamic stereo matching. Unlike the existing methods, we model this task as local matching and global aggregation. Locally, we consider correlation in a triple-frame manner to pool information from adjacent frames and improve the temporal consistency. Globally, to exploit the entire sequence's consistency and extract dynamic scene cues for aggregation, we develop a motion-propagation recurrent unit. Extensive experiments demonstrate the performance of our method, showcasing improvements in prediction quality and achieving state-of-the-art results on various commonly used benchmarks.
Accurate depth and semantic segmentation are crucial for various computer vision tasks. However, the scarcity of annotated real-world aerial datasets poses a significant challenge for training and evaluating robust models. Additionally, the detection and segmentation of thin objects, such as wires, cables, and fences, present a critical concern for ensuring the safe operation of drones. To address these limitations, we present a novel synthetic dataset specifically designed for depth and semantic segmentation tasks in aerial views. Leveraging photo-realistic rendering techniques, our dataset provides a valuable resource for training models using a synthetic-supervision training scheme while introducing new drone-specific metrics for depth accuracy.
Collaborative inference systems are one of the emerging solutions for deploying deep neural networks (DNNs) at the wireless network edge. Their main idea is to divide a DNN into two parts, where the first is shallow enough to be reliably executed at edge devices of limited computational power, while the second part is executed at an edge server with higher computational capabilities. The main advantage of such systems is that the input of the DNN gets compressed as the subsequent layers of the shallow part extract only the information necessary for the task. As a result, significant communication savings can be achieved compared to transmitting raw input samples. In this work, we study early exiting in the context of collaborative inference, which allows obtaining inference results at the edge device for certain samples, without the need to transmit the partially processed data to the edge server at all, leading to further communication savings. The central part of our system is the transmission-decision (TD) mechanism, which, given the information from the early exit, and the wireless channel conditions, decides whether to keep the early exit prediction or transmit the data to the edge server for further processing. In this paper, we evaluate various TD mechanisms and show experimentally, that for an image classification task over the wireless edge, proper utilization of early exits can provide both performance gains and significant communication savings.
This paper introduces a vision transformer (ViT)-based deep joint source and channel coding (DeepJSCC) scheme for wireless image transmission over multiple-input multiple-output (MIMO) channels, denoted as DeepJSCC-MIMO. We consider DeepJSCC-MIMO for adaptive image transmission in both open-loop and closed-loop MIMO systems. The novel DeepJSCC-MIMO architecture surpasses the classical separation-based benchmarks with robustness to channel estimation errors and showcases remarkable flexibility in adapting to diverse channel conditions and antenna numbers without requiring retraining. Specifically, by harnessing the self-attention mechanism of ViT, DeepJSCC-MIMO intelligently learns feature mapping and power allocation strategies tailored to the unique characteristics of the source image and prevailing channel conditions. Extensive numerical experiments validate the significant improvements in transmission quality achieved by DeepJSCC-MIMO for both open-loop and closed-loop MIMO systems across a wide range of scenarios. Moreover, DeepJSCC-MIMO exhibits robustness to varying channel conditions, channel estimation errors, and different antenna numbers, making it an appealing solution for emerging semantic communication systems.
This paper presents a novel wireless image transmission paradigm that can exploit feedback from the receiver, called DeepJSCC-ViT-f. We consider a block feedback channel model, where the transmitter receives noiseless/noisy channel output feedback after each block. The proposed scheme employs a single encoder to facilitate transmission over multiple blocks, refining the receiver's estimation at each block. Specifically, the unified encoder of DeepJSCC-ViT-f can leverage the semantic information from the source image, and acquire channel state information and the decoder's current belief about the source image from the feedback signal to generate coded symbols at each block. Numerical experiments show that our DeepJSCC-ViT-f scheme achieves state-of-the-art transmission performance with robustness to noise in the feedback link. Additionally, DeepJSCC-ViT-f can adapt to the channel condition directly through feedback without the need for separate channel estimation. We further extend the scope of the DeepJSCC-ViT-f approach to include the broadcast channel, which enables the transmitter to generate broadcast codes in accordance with signal semantics and channel feedback from individual receivers.
We study the collaborative image retrieval problem at the wireless edge, where multiple edge devices capture images of the same object, which are then used jointly to retrieve similar images at the edge server over a shared multiple access channel. We propose a semantic non-orthogonal multiple access (NOMA) communication paradigm, in which extracted features from each device are mapped directly to channel inputs, which are then added over-the-air. We propose a novel contrastive learning (CL)-based semantic communication (CL-SC) paradigm, aiming to exploit signal correlations to maximize the retrieval accuracy under a total bandwidth constraints. Specifically, we treat noisy correlated signals as different augmentations of a common identity, and propose a cross-view CL algorithm to optimize the correlated signals in a coarse-to-fine fashion to improve retrieval accuracy. Extensive numerical experiments verify that our method achieves the state-of-the-art performance and can significantly improve retrieval accuracy, with particularly significant gains in low signla-to-noise ratio (SNR) and limited bandwidth regimes.