Recently, heatmap regression models have become the mainstream in locating facial landmarks. To keep computation affordable and reduce memory usage, the whole procedure involves downsampling from the raw image to the output heatmap. However, how much impact will the quantization error introduced by downsampling bring? The problem is hardly systematically investigated among previous works. This work fills the blank and we are the first to quantitatively analyze the negative gain. The statistical results show the NME generated by quantization error is even larger than 1/3 of the SOTA item, which is a serious obstacle for making a new breakthrough in face alignment. To compensate the impact of quantization effect, we propose a novel method, called Heatmap In Heatmap(HIH), which leverages two categories of heatmaps as label representation to encode coordinate. And in HIH, the range of one heatmap represents a pixel of the other category of heatmap. Also, we even combine the face alignment with solutions of other fields to make a comparison. Extensive experiments on various benchmarks show the feasibility of HIH and the superior performance than other solutions. Moreover, the mean error reaches to 4.18 on WFLW, which exceeds SOTA a lot. Our source code are made publicly available at supplementary material.
Existing methods for skeleton-based action recognition mainly focus on improving the recognition accuracy, whereas the efficiency of the model is rarely considered. Recently, there are some works trying to speed up the skeleton modeling by designing light-weight modules. However, in addition to the model size, the amount of the data involved in the calculation is also an important factor for the running speed, especially for the skeleton data where most of the joints are redundant or non-informative to identify a specific skeleton. Besides, previous works usually employ one fix-sized model for all the samples regardless of the difficulty of recognition, which wastes computations for easy samples. To address these limitations, a novel approach, called AdaSGN, is proposed in this paper, which can reduce the computational cost of the inference process by adaptively controlling the input number of the joints of the skeleton on-the-fly. Moreover, it can also adaptively select the optimal model size for each sample to achieve a better trade-off between accuracy and efficiency. We conduct extensive experiments on three challenging datasets, namely, NTU-60, NTU-120 and SHREC, to verify the superiority of the proposed approach, where AdaSGN achieves comparable or even higher performance with much lower GFLOPs compared with the baseline method.
This paper revisits feature pyramids networks (FPN) for one-stage detectors and points out that the success of FPN is due to its divide-and-conquer solution to the optimization problem in object detection rather than multi-scale feature fusion. From the perspective of optimization, we introduce an alternative way to address the problem instead of adopting the complex feature pyramids - {\em utilizing only one-level feature for detection}. Based on the simple and efficient solution, we present You Only Look One-level Feature (YOLOF). In our method, two key components, Dilated Encoder and Uniform Matching, are proposed and bring considerable improvements. Extensive experiments on the COCO benchmark prove the effectiveness of the proposed model. Our YOLOF achieves comparable results with its feature pyramids counterpart RetinaNet while being $2.5\times$ faster. Without transformer layers, YOLOF can match the performance of DETR in a single-level feature manner with $7\times$ less training epochs. With an image size of $608\times608$, YOLOF achieves 44.3 mAP running at 60 fps on 2080Ti, which is $13\%$ faster than YOLOv4. Code is available at \url{https://github.com/megvii-model/YOLOF}.
BERT is the most recent Transformer-based model that achieves state-of-the-art performance in various NLP tasks. In this paper, we investigate the hardware acceleration of BERT on FPGA for edge computing. To tackle the issue of huge computational complexity and memory footprint, we propose to fully quantize the BERT (FQ-BERT), including weights, activations, softmax, layer normalization, and all the intermediate results. Experiments demonstrate that the FQ-BERT can achieve 7.94x compression for weights with negligible performance loss. We then propose an accelerator tailored for the FQ-BERT and evaluate on Xilinx ZCU102 and ZCU111 FPGA. It can achieve a performance-per-watt of 3.18 fps/W, which is 28.91x and 12.72x over Intel(R) Core(TM) i7-8700 CPU and NVIDIA K80 GPU, respectively.
Convolutional neural networks are able to learn realistic image priors from numerous training samples in low-level image generation and restoration. We show that, for high-level image recognition tasks, we can further reconstruct "realistic" images of each category by leveraging intrinsic Batch Normalization (BN) statistics without any training data. Inspired by the popular VAE/GAN methods, we regard the zero-shot optimization process of synthetic images as generative modeling to match the distribution of BN statistics. The generated images serve as a calibration set for the following zero-shot network quantizations. Our method meets the needs for quantizing models based on sensitive information, \textit{e.g.,} due to privacy concerns, no data is available. Extensive experiments on benchmark datasets show that, with the help of generated data, our approach consistently outperforms existing data-free quantization methods.
Time series classification problems exist in many fields and have been explored for a couple of decades. However, they still remain challenging, and their solutions need to be further improved for real-world applications in terms of both accuracy and efficiency. In this paper, we propose a hybrid neural architecture, called Self-Attentive Recurrent Convolutional Networks (SARCoN), to learn multi-faceted representations for univariate time series. SARCoN is the synthesis of long short-term memory networks with self-attentive mechanisms and Fully Convolutional Networks, which work in parallel to learn the representations of univariate time series from different perspectives. The component modules of the proposed architecture are trained jointly in an end-to-end manner and they classify the input time series in a cooperative way. Due to its domain-agnostic nature, SARCoN is able to generalize a diversity of domain tasks. Our experimental results show that, compared to the state-of-the-art approaches for time series classification, the proposed architecture can achieve remarkable improvements for a set of univariate time series benchmarks from the UCR repository. Moreover, the self-attention and the global average pooling in the proposed architecture enable visible interpretability by facilitating the identification of the contribution regions of the original time series. An overall analysis confirms that multi-faceted representations of time series aid in capturing deep temporal corrections within complex time series, which is essential for the improvement of time series classification performance. Our work provides a novel angle that deepens the understanding of time series classification, qualifying our proposed model as an ideal choice for real-world applications.
This paper reviews the second AIM realistic bokeh effect rendering challenge and provides the description of the proposed solutions and results. The participating teams were solving a real-world bokeh simulation problem, where the goal was to learn a realistic shallow focus technique using a large-scale EBB! bokeh dataset consisting of 5K shallow / wide depth-of-field image pairs captured using the Canon 7D DSLR camera. The participants had to render bokeh effect based on only one single frame without any additional data from other cameras or sensors. The target metric used in this challenge combined the runtime and the perceptual quality of the solutions measured in the user study. To ensure the efficiency of the submitted models, we measured their runtime on standard desktop CPUs as well as were running the models on smartphone GPUs. The proposed solutions significantly improved the baseline results, defining the state-of-the-art for practical bokeh effect rendering problem.
A photo captured with bokeh effect often means objects in focus are sharp while the out-of-focus areas are all blurred. DSLR can easily render this kind of effect naturally. However, due to the limitation of sensors, smartphones cannot capture images with depth-of-field effects directly. In this paper, we propose a novel generator called Glass-Net, which generates bokeh images not relying on complex hardware. Meanwhile, the GAN-based method and perceptual loss are combined for rendering a realistic bokeh effect in the stage of finetuning the model. Moreover, Instance Normalization(IN) is reimplemented in our network, which ensures our tflite model with IN can be accelerated on smartphone GPU. Experiments show that our method is able to render a high-quality bokeh effect and process one $1024 \times 1536$ pixel image in 1.9 seconds on all smartphone chipsets. This approach ranked First in AIM 2020 Rendering Realistic Bokeh Challenge Track 1 \& Track 2.
Human identification is an important topic in event detection, person tracking, and public security. There have been numerous methods proposed for human identification, such as face identification, person re-identification, and gait identification. Typically, existing methods predominantly classify a queried image to a specific identity in an image gallery set (I2I). This is seriously limited for the scenario where only a textual description of the query or an attribute gallery set is available in a wide range of video surveillance applications (A2I or I2A). However, very few efforts have been devoted towards modality-free identification, i.e., identifying a query in a gallery set in a scalable way. In this work, we take an initial attempt, and formulate such a novel Modality-Free Human Identification (named MFHI) task as a generic zero-shot learning model in a scalable way. Meanwhile, it is capable of bridging the visual and semantic modalities by learning a discriminative prototype of each identity. In addition, the semantics-guided spatial attention is enforced on visual modality to obtain representations with both high global category-level and local attribute-level discrimination. Finally, we design and conduct an extensive group of experiments on two common challenging identification tasks, including face identification and person re-identification, demonstrating that our method outperforms a wide variety of state-of-the-art methods on modality-free human identification.
We review the AIM 2020 challenge on virtual image relighting and illumination estimation. This paper presents the novel VIDIT dataset used in the challenge and the different proposed solutions and final evaluation results over the 3 challenge tracks. The first track considered one-to-one relighting; the objective was to relight an input photo of a scene with a different color temperature and illuminant orientation (i.e., light source position). The goal of the second track was to estimate illumination settings, namely the color temperature and orientation, from a given image. Lastly, the third track dealt with any-to-any relighting, thus a generalization of the first track. The target color temperature and orientation, rather than being pre-determined, are instead given by a guide image. Participants were allowed to make use of their track 1 and 2 solutions for track 3. The tracks had 94, 52, and 56 registered participants, respectively, leading to 20 confirmed submissions in the final competition stage.