다중혈관 관상동맥 환자에서 y-문합을 이용하여 양쪽 내흉동맥만을 사용한 우회술의 조기 성적

“Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks”の学習報告

Recently, Neural Architecture Search (NAS) has successfully identified neural network architectures that exceed human designed ones on large-scale image classification. In this paper, we study NAS for semantic image segmentation. Existing works often focus on searching the repeatable cell structure, while hand-designing the outer network structure that controls the spatial resolution changes. This choice simplifies the search space, but becomes increasingly problematic for dense image prediction which exhibits a lot more network level architectural variations. Therefore, we propose to search the network level structure in addition to the cell level structure, which forms a hierarchical architecture search space. We present a network level search space that includes many popular designs, and develop a formulation that allows efficient gradient-based architecture search (3 P100 GPU days on Cityscapes images). We demonstrate the effectiveness of the proposed method on the challenging Cityscapes, PASCAL VOC 2012, and ADE20K datasets. Auto-DeepLab, our architecture searched specifically for semantic image segmentation, attains state-of-the-art performance without any ImageNet pretraining.

/pdf/auto-deeplab-hierarchical-neural-architecture-search-for-rsvt67m5s2.pdf

Auto-DeepLab: Hierarchical Neural Architecture Search for Semantic Image Segmentation

Person re-identification (Re-ID) aims at retrieving a person of interest across multiple non-overlapping cameras. With the advancement of deep neural networks and increasing demand of intelligent video surveillance, it has gained significantly increased interest in the computer vision community. By dissecting the involved components in developing a person Re-ID system, we categorize it into the closed-world and open-world settings. The widely studied closed-world setting is usually applied under various research-oriented assumptions, and has achieved inspiring success using deep learning techniques on a number of datasets. We first conduct a comprehensive overview with in-depth analysis for closed-world person Re-ID from three different perspectives, including deep feature representation learning, deep metric learning and ranking optimization. With the performance saturation under closed-world setting, the research focus for person Re-ID has recently shifted to the open-world setting, facing more challenging issues. This setting is closer to practical applications under specific scenarios. We summarize the open-world Re-ID in terms of five different aspects. By analyzing the advantages of existing methods, we design a powerful AGW baseline, achieving state-of-the-art or at least comparable performance on twelve datasets for FOUR different Re-ID tasks. Meanwhile, we introduce a new evaluation metric (mINP) for person Re-ID, indicating the cost for finding all the correct matches, which provides an additional criteria to evaluate the Re-ID system for real applications. Finally, some important yet under-investigated open issues are discussed.

Deep Learning for Person Re-identification: A Survey and Outlook

Recent advancements in perception for autonomous driving are driven by deep learning. In order to achieve robust and accurate scene understanding, autonomous vehicles are usually equipped with different sensors (e.g. cameras, LiDARs, Radars), and multiple sensing modalities can be fused to exploit their complementary properties. In this context, many methods have been proposed for deep multi-modal perception problems. However, there is no general guideline for network architecture design, and questions of “what to fuse”, “when to fuse”, and “how to fuse” remain open. This review paper attempts to systematically summarize methodologies and discuss challenges for deep multi-modal object detection and semantic segmentation in autonomous driving. To this end, we first provide an overview of on-board sensors on test vehicles, open datasets, and background information for object detection and semantic segmentation in autonomous driving research. We then summarize the fusion methodologies and discuss challenges and open questions. In the appendix, we provide tables that summarize topics and methods. We also provide an interactive online platform to navigate each reference: https://boschresearch.github.io/multimodalperception/ .

Deep Multi-Modal Object Detection and Semantic Segmentation for Autonomous Driving: Datasets, Methods, and Challenges

Like H.264, AVS video coding standard also uses macroblock (MB) based motion compensation (MC) and mode decision (MD). Rate distortion optimization (RDO) is the best known mode decision method, but with a high computational complexity that limits its applications. In our paper, firstly an MD algorithm based on RDO is given, which makes more mode candidates enter into RDO mode decision with little hardware resource increment. We further analyze the pipeline structure in detail, and implement a block-level 5-stage hardware pipeline. It can support the real time RDO mode decision processing of 1080P@30fps, and the coding efficiency is about 0.5db higher than the traditional SAD method. Our design is described in high-level Verilog/VHDL hardware description language and implemented under SMIC 0.18-µm CMOS technology with 215K logic gates and 80 KB SRAMs.

A highly efficient pipeline architecture of RDO-based mode decision design for AVS HD video encoder

Integer motion estimation (IME), which acts as a key component in video encoder, is to remove temporal redundancies by searching the best integer motion vectors for dynamic partition blocks in a macro-block (MB). Huge memory bandwidth requirements and unbearable computational resource demanding are two key bottlenecks in IME engine design, especially for large search window (SW) cases. In this paper, a three-level pipelined VLSI architecture design is proposed, where efficiently integrates the reference data sharing search (RDSS) into multi-resolution motion estimation algorithm (MMEA). First, a hardware-friendly MMEA algorithm is mapped into three-level pipelined architecture with neglected coding quality loss. Second, sub-sampled RDSS coupled with Level C + are adopted to reduce on-chip memory and bandwidth at the coarsest and middle level. Data sharing between IME and fractional motion estimation (FME) is achieved by loading only a local predictive SW at the finest level. Finally, the three levels are parallelized and pipelined to guarantee the gradual refinement of MMEA and the hardware utilization. Experimental results show that the proposed architecture can reach a good balance among complexity, on-chip memory, bandwidth, and the data flow regularity. Only 320 processing elements (PE) within 550 cycles are required for IME search, where the SW is set to 256 × 256. Our architecture can achieve 1080P@30 fps real-time processing at the working frequency of 134.6 MHz, with 135 K gates and 8.93 KB on-chip memory.

Three-level pipelined multi-resolution integer motion estimation engine with optimized reference data sharing search for AVS

Video segmentation aims to segment target objects in a video sequence, which remains a challenge due to the motion and deformation of objects. In this paper, we propose a novel attention-driven hybrid encoder-decoder network that generates object segmentation by fully leveraging spatial and temporal information. Firstly, a multi-branch network is designed to learn feature representation from object appearance, location and motion. Secondly, a target attention module is proposed to further exploit context information from learned representation. In addition, a novel edge loss is designed which constraints the model to generate salient edge features and accurate segmentation. The proposed model has been evaluated over two widely used public benchmarks, and experiments demonstrate its superior robustness and effectiveness as compared with the state of the arts.

Fusion Target Attention Mask Generation Network For Video Segmentation

In this paper, we propose a novel image coding and transmission scheme called AnalogCast, which is a pseudo analog coding system for transmitting satellite remote-sensing images to large number of receivers. AnalogCast follows the idea originally developed for SoftCast [1–3] but with two special techniques, scale factor estimate based on L-shaped chunk division and scale factor fix curve-fitting model, so that there is no need to transmit side information as is required for SoftCast [1–3]. This improvement can get better mobility, less bandwidth and computational budget than SoftCast [1–3]. For robustness, AnalogCast adopts a mode of GOP (Group of Pictures) interweave structure to smooth image quality and improve perceptual quality when packet loss happens. Experimental results show that the proposed method can work well in low SNR condition in −7dB to 4dB where SoftCast system and JPEG2000 system fall into "cliff effect". And in high SNR condition, our proposed method works even better than SoftCast system at SNR 1.5∼3dB and a gain of 1∼10dB over JPEG2000 with forward error correction. For robustness, the AnalogCast works better than SoftCast at SNR 8dB when the burst packet loss rate is higher than 35%. And the scheme gets better perceptual quality under packet loss.

AnalogCast: Full linear coding and pseudo analog transmission for satellite remote-sensing images

Search range (SR) is a key parameter on the search quality control for motion estimation (ME) of a real-time video encoder. Dynamic search range (DSR) is a commonly employed algorithm to reduce the computational complexity of ME in a video encoder. In this paper, we model an effective predicted motion vector (PMV) deviation metric to predict the relationship between SR and motion vector difference (MVD), according to the prediction differences of both temporal and spatial motions of neighboring blocks. In addition, a computation-constrained DSR (CDSR) control algorithm is proposed to manage the computational complexity while maximizing video coding quality in a real-time computational constrained scenario. The SR is dynamically determined by three factors: motion complexity, user-defined probability and computation budget. Compared to the conventional DSR algorithms, the proposed CDSR is an effective and quantifiable algorithm to allocate more computation budget to the blocks with high PMV deviations (such as motion object boundary), and less computation budget to the well-matched motion predicted blocks, while maintaining a constrained computation requirement. Experimental results show that the proposed CDSR control algorithm is an effective method to manage the computation consumption of the DSR algorithm while keeping similar rate-distortion (RD) performance. It can achieve about 0.1-0.3dB average PSNR improvement when the computation consumption is restricted to a specific level as compared with its equivalent Fixed SR algorithm and can achieve about 50-90% computation savings when compared to the benchmarks. For ME with high performance Processing Element (PE) engine, the quality degradation caused by the proposed CDSR algorithm can be ignored. Proposed an effective metric to model the relationship between SR and MVD.Proposed a linear model for computational complexity prediction.Proposed a quantifiable computation-constrained DSR control algorithm.

Huizhu Jia

Papers

A highly efficient pipeline architecture of RDO-based mode decision design for AVS HD video encoder

Three-level pipelined multi-resolution integer motion estimation engine with optimized reference data sharing search for AVS

Fusion Target Attention Mask Generation Network For Video Segmentation

AnalogCast: Full linear coding and pseudo analog transmission for satellite remote-sensing images

Computation-constrained dynamic search range control for real-time video encoder