Xu Jia

Haiwen Diao, Bo Wan, Xu Jia et al.

Parameter-efficient transfer learning (PETL) has emerged as a flourishing research field for adapting large pre-trained models to downstream tasks, greatly reducing trainable parameters while grappling with memory challenges during fine-tuning. To address it, memory-efficient series (METL) avoid backpropagating gradients through the large backbone. However, they compromise by exclusively relying on frozen intermediate outputs and limiting the exhaustive exploration of prior knowledge from pre-trained models. Moreover, the dependency and redundancy between cross-layer features are frequently overlooked, thereby submerging more discriminative representations and causing an inherent performance gap (vs. conventional PETL methods). Hence, we propose an innovative METL strategy called SHERL for resource-limited scenarios to decouple the entire adaptation into two successive and complementary processes. In the early route, intermediate outputs are consolidated via an anti-redundancy operation, enhancing their compatibility for subsequent interactions; thereby in the late route, utilizing minimal late pre-trained layers could alleviate the peak demand on memory overhead and regulate these fairly flexible features into more adaptive and powerful representations for new domains. Extensive ablations on vision-and-language and language-only tasks show that SHERL combines the strengths of both parameter and memory-efficient techniques, performing on-par or better across diverse architectures with lower memory during fine-tuning. Our code is publicly available at: https://github.com/Paranioar/SHERL.

Xuejun Li, Ran Ding, Xiao Liu et al.

Current cloud-based smart systems suffer from weaknesses such as high response latency, limited network bandwidth and the restricted computing power of smart end devices which seriously affect the system's QoS (Quality of Service). Recently, given its advantages of low latency, high bandwidth and location awareness, edge computing has become a promising solution for smart systems. However, the development of edge computing based smart systems is a very challenging job for software developers who do not have the skills for the creation of edge computing environments. The management of edge computing resources and computing tasks is also very challenging. Workflow technology has been widely used in smart systems to automate task and resource management, but there does not yet exist a real-world deployable edge computing based workflow execution engine. To fill this gap, we present EdgeWorkflowReal, an edge computing based workflow execution engine for smart systems. EdgeWorkflowReal supports: 1) automatic creation of a real edge computing environment according to user settings; 2) visualized modelling of edge workflow applications; and 3) automatic deployment, monitoring and performance evaluation of edge workflow applications in a smart system.

Pengpeng Liu, Irwin King, Michael R. Lyu et al.

We present DDFlow, a data distillation approach to learning optical flow estimation from unlabeled data. The approach distills reliable predictions from a teacher network, and uses these predictions as annotations to guide a student network to learn optical flow. Unlike existing work relying on hand-crafted energy terms to handle occlusion, our approach is data-driven, and learns optical flow for occluded pixels. This enables us to train our model with a much simpler loss function, and achieve a much higher accuracy. We conduct a rigorous evaluation on the challenging Flying Chairs, MPI Sintel, KITTI 2012 and 2015 benchmarks, and show that our approach significantly outperforms all existing unsupervised learning methods, while running at real time.

Liqian Ma, Xu Jia, Stamatios Georgoulis et al.

Image-to-image translation has recently received significant attention due to advances in deep learning. Most works focus on learning either a one-to-one mapping in an unsupervised way or a many-to-many mapping in a supervised way. However, a more practical setting is many-to-many mapping in an unsupervised way, which is harder due to the lack of supervision and the complex inner- and cross-domain variations. To alleviate these issues, we propose the Exemplar Guided & Semantically Consistent Image-to-image Translation (EGSC-IT) network which conditions the translation process on an exemplar image in the target domain. We assume that an image comprises of a content component which is shared across domains, and a style component specific to each domain. Under the guidance of an exemplar from the target domain we apply Adaptive Instance Normalization to the shared content component, which allows us to transfer the style information of the target domain to the source domain. To avoid semantic inconsistencies during translation that naturally appear due to the large inner- and cross-domain variations, we introduce the concept of feature masks that provide coarse semantic guidance without requiring the use of any semantic labels. Experimental results on various datasets show that EGSC-IT does not only translate the source image to diverse instances in the target domain, but also preserves the semantic consistency during the process.

Liqian Ma, Xu Jia, Qianru Sun et al.

This paper proposes the novel Pose Guided Person Generation Network (PG$^2$) that allows to synthesize person images in arbitrary poses, based on an image of that person and a novel pose. Our generation framework PG$^2$ utilizes the pose information explicitly and consists of two key stages: pose integration and image refinement. In the first stage the condition image and the target pose are fed into a U-Net-like network to generate an initial but coarse image of the person with the target pose. The second stage then refines the initial and blurry result by training a U-Net-like generator in an adversarial way. Extensive experimental results on both 128$\times$64 re-identification images and 256$\times$256 fashion photos show that our model generates high-quality person images with convincing details.

Jia Xu, René Ranftl, Vladlen Koltun

We present an optical flow estimation approach that operates on the full four-dimensional cost volume. This direct approach shares the structural benefits of leading stereo matching pipelines, which are known to yield high accuracy. To this day, such approaches have been considered impractical due to the size of the cost volume. We show that the full four-dimensional cost volume can be constructed in a fraction of a second due to its regularity. We then exploit this regularity further by adapting semi-global matching to the four-dimensional setting. This yields a pipeline that achieves significantly higher accuracy than state-of-the-art optical flow methods while being faster than most. Our approach outperforms all published general-purpose optical flow methods on both Sintel and KITTI 2015 benchmarks.

Amir Ghodrati, Xu Jia, Marco Pedersoli et al.

Learning the distribution of images in order to generate new samples is a challenging task due to the high dimensionality of the data and the highly non-linear relations that are involved. Nevertheless, some promising results have been reported in the literature recently,building on deep network architectures. In this work, we zoom in on a specific type of image generation: given an image and knowing the category of objects it belongs to (e.g. faces), our goal is to generate a similar and plausible image, but with some altered attributes. This is particularly challenging, as the model needs to learn to disentangle the effect of each attribute and to apply a desired attribute change to a given input image, while keeping the other attributes and overall object appearance intact. To this end, we learn a convolutional network, where the desired attribute information is encoded then merged with the encoded image at feature map level. We show promising results, both qualitatively as well as quantitatively, in the context of a retrieval experiment, on two face datasets (MultiPie and CAS-PEAL-R1).

Xu Jia, Efstratios Gavves, Basura Fernando et al.

In this work we focus on the problem of image caption generation. We propose an extension of the long short term memory (LSTM) model, which we coin gLSTM for short. In particular, we add semantic information extracted from the image as extra input to each unit of the LSTM block, with the aim of guiding the model towards solutions that are more tightly coupled to the image content. Additionally, we explore different length normalization strategies for beam search in order to prevent from favoring short sentences. On various benchmark datasets such as Flickr8K, Flickr30K and MS COCO, we obtain results that are on par with or even outperform the current state-of-the-art.