In this research work, we have demonstrated the application of Mask-RCNN (Regional Convolutional Neural Network), a deep-learning algorithm for computer vision and specifically object detection, to semiconductor defect inspection domain. Stochastic defect detection and classification during semiconductor manufacturing has grown to be a challenging task as we continuously shrink circuit pattern dimensions (e.g., for pitches less than 32 nm). Defect inspection and analysis by state-of-the-art optical and e-beam inspection tools is generally driven by some rule-based techniques, which in turn often causes to misclassification and thereby necessitating human expert intervention. In this work, we have revisited and extended our previous deep learning-based defect classification and detection method towards improved defect instance segmentation in SEM images with precise extent of defect as well as generating a mask for each defect category/instance. This also enables to extract and calibrate each segmented mask and quantify the pixels that make up each mask, which in turn enables us to count each categorical defect instances as well as to calculate the surface area in terms of pixels. We are aiming at detecting and segmenting different types of inter-class stochastic defect patterns such as bridge, break, and line collapse as well as to differentiate accurately between intra-class multi-categorical defect bridge scenarios (as thin/single/multi-line/horizontal/non-horizontal) for aggressive pitches as well as thin resists (High NA applications). Our proposed approach demonstrates its effectiveness both quantitatively and qualitatively.
The global Information and Communications Technology (ICT) supply chain is a complex network consisting of all types of participants. It is often formulated as a Social Network to discuss the supply chain network's relations, properties, and development in supply chain management. Information sharing plays a crucial role in improving the efficiency of the supply chain, and datasheets are the most common data format to describe e-component commodities in the ICT supply chain because of human readability. However, with the surging number of electronic documents, it has been far beyond the capacity of human readers, and it is also challenging to process tabular data automatically because of the complex table structures and heterogeneous layouts. Table Structure Recognition (TSR) aims to represent tables with complex structures in a machine-interpretable format so that the tabular data can be processed automatically. In this paper, we formulate TSR as an object detection problem and propose to generate an intuitive representation of a complex table structure to enable structuring of the tabular data related to the commodities. To cope with border-less and small layouts, we propose a cost-sensitive loss function by considering the detection difficulty of each class. Besides, we propose a novel anchor generation method using the character of tables that columns in a table should share an identical height, and rows in a table should share the same width. We implement our proposed method based on Faster-RCNN and achieve 94.79% on mean Average Precision (AP), and consistently improve more than 1.5% AP for different benchmark models.
There is a trend to fuse multi-modal information for 3D object detection (3OD). However, the challenging problems of low lightweightness, poor flexibility of plug-and-play, and inaccurate alignment of features are still not well-solved, when designing multi-modal fusion newtorks. We propose PointSee, a lightweight, flexible and effective multi-modal fusion solution to facilitate various 3OD networks by semantic feature enhancement of LiDAR point clouds assembled with scene images. Beyond the existing wisdom of 3OD, PointSee consists of a hidden module (HM) and a seen module (SM): HM decorates LiDAR point clouds using 2D image information in an offline fusion manner, leading to minimal or even no adaptations of existing 3OD networks; SM further enriches the LiDAR point clouds by acquiring point-wise representative semantic features, leading to enhanced performance of existing 3OD networks. Besides the new architecture of PointSee, we propose a simple yet efficient training strategy, to ease the potential inaccurate regressions of 2D object detection networks. Extensive experiments on the popular outdoor/indoor benchmarks show numerical improvements of our PointSee over twenty-two state-of-the-arts.
The ground plane prior is a very informative geometry clue in monocular 3D object detection (M3OD). However, it has been neglected by most mainstream methods. In this paper, we identify two key factors that limit the applicability of ground plane prior: the projection point localization issue and the ground plane tilt issue. To pick up the ground plane prior for M3OD, we propose a Ground Plane Enhanced Network (GPENet) which resolves both issues at one go. For the projection point localization issue, instead of using the bottom vertices or bottom center of the 3D bounding box (BBox), we leverage the object's ground contact points, which are explicit pixels in the image and easy for the neural network to detect. For the ground plane tilt problem, our GPENet estimates the horizon line in the image and derives a novel mathematical expression to accurately estimate the ground plane equation. An unsupervised vertical edge mining algorithm is also proposed to address the occlusion of the horizon line. Furthermore, we design a novel 3D bounding box deduction method based on a dynamic back projection algorithm, which could take advantage of the accurate contact points and the ground plane equation. Additionally, using only M3OD labels, contact point and horizon line pseudo labels can be easily generated with NO extra data collection and label annotation cost. Extensive experiments on the popular KITTI benchmark show that our GPENet can outperform other methods and achieve state-of-the-art performance, well demonstrating the effectiveness and the superiority of the proposed approach. Moreover, our GPENet works better than other methods in cross-dataset evaluation on the nuScenes dataset. Our code and models will be published.
Frozen pretrained models have become a viable alternative to the pretraining-then-finetuning paradigm for transfer learning. However, with frozen models there are relatively few parameters available for adapting to downstream tasks, which is problematic in computer vision where tasks vary significantly in input/output format and the type of information that is of value. In this paper, we present a study of frozen pretrained models when applied to diverse and representative computer vision tasks, including object detection, semantic segmentation and video action recognition. From this empirical analysis, our work answers the questions of what pretraining task fits best with this frozen setting, how to make the frozen setting more flexible to various downstream tasks, and the effect of larger model sizes. We additionally examine the upper bound of performance using a giant frozen pretrained model with 3 billion parameters (SwinV2-G) and find that it reaches competitive performance on a varied set of major benchmarks with only one shared frozen base network: 60.0 box mAP and 52.2 mask mAP on COCO object detection test-dev, 57.6 val mIoU on ADE20K semantic segmentation, and 81.7 top-1 accuracy on Kinetics-400 action recognition. With this work, we hope to bring greater attention to this promising path of freezing pretrained image models.
Recently, the dominant DETR-based approaches apply central-concept spatial prior to accelerate Transformer detector convergency. These methods gradually refine the reference points to the center of target objects and imbue object queries with the updated central reference information for spatially conditional attention. However, centralizing reference points may severely deteriorate queries' saliency and confuse detectors due to the indiscriminative spatial prior. To bridge the gap between the reference points of salient queries and Transformer detectors, we propose SAlient Point-based DETR (SAP-DETR) by treating object detection as a transformation from salient points to instance objects. In SAP-DETR, we explicitly initialize a query-specific reference point for each object query, gradually aggregate them into an instance object, and then predict the distance from each side of the bounding box to these points. By rapidly attending to query-specific reference region and other conditional extreme regions from the image features, SAP-DETR can effectively bridge the gap between the salient point and the query-based Transformer detector with a significant convergency speed. Our extensive experiments have demonstrated that SAP-DETR achieves 1.4 times convergency speed with competitive performance. Under the standard training scheme, SAP-DETR stably promotes the SOTA approaches by 1.0 AP. Based on ResNet-DC-101, SAP-DETR achieves 46.9 AP.
Few-Shot Learning (FSL) alleviates the data shortage challenge via embedding discriminative target-aware features among plenty seen (base) and few unseen (novel) labeled samples. Most feature embedding modules in recent FSL methods are specially designed for corresponding learning tasks (e.g., classification, segmentation, and object detection), which limits the utility of embedding features. To this end, we propose a light and universal module named transformer-based Semantic Filter (tSF), which can be applied for different FSL tasks. The proposed tSF redesigns the inputs of a transformer-based structure by a semantic filter, which not only embeds the knowledge from whole base set to novel set but also filters semantic features for target category. Furthermore, the parameters of tSF is equal to half of a standard transformer block (less than 1M). In the experiments, our tSF is able to boost the performances in different classic few-shot learning tasks (about 2% improvement), especially outperforms the state-of-the-arts on multiple benchmark datasets in few-shot classification task.
Inspired by the success of vision-language methods (VLMs) in zero-shot classification, recent works attempt to extend this line of work into object detection by leveraging the localization ability of pre-trained VLMs and generating pseudo labels for unseen classes in a self-training manner. However, since the current VLMs are usually pre-trained with aligning sentence embedding with global image embedding, the direct use of them lacks fine-grained alignment for object instances, which is the core of detection. In this paper, we propose a simple but effective fine-grained Visual-Text Prompt-driven self-training paradigm for Open-Vocabulary Detection (VTP-OVD) that introduces a fine-grained visual-text prompt adapting stage to enhance the current self-training paradigm with a more powerful fine-grained alignment. During the adapting stage, we enable VLM to obtain fine-grained alignment by using learnable text prompts to resolve an auxiliary dense pixel-wise prediction task. Furthermore, we propose a visual prompt module to provide the prior task information (i.e., the categories need to be predicted) for the vision branch to better adapt the pre-trained VLM to the downstream tasks. Experiments show that our method achieves the state-of-the-art performance for open-vocabulary object detection, e.g., 31.5% mAP on unseen classes of COCO.
We propose a method for joint detection and tracking of multiple objects in 3D point clouds, a task conventionally treated as a two-step process comprising object detection followed by data association. Our method embeds both steps into a single end-to-end trainable network eliminating the dependency on external object detectors. Our model exploits temporal information employing multiple frames to detect objects and track them in a single network, thereby making it a utilitarian formulation for real-world scenarios. Computing affinity matrix by employing features similarity across consecutive point cloud scans forms an integral part of visual tracking. We propose an attention-based refinement module to refine the affinity matrix by suppressing erroneous correspondences. The module is designed to capture the global context in affinity matrix by employing self-attention within each affinity matrix and cross-attention across a pair of affinity matrices. Unlike competing approaches, our network does not require complex post-processing algorithms, and processes raw LiDAR frames to directly output tracking results. We demonstrate the effectiveness of our method on the three tracking benchmarks: JRDB, Waymo, and KITTI. Experimental evaluations indicate the ability of our model to generalize well across datasets.
Circle representation has recently been introduced as a medical imaging optimized representation for more effective instance object detection on ball-shaped medical objects. With its superior performance on instance detection, it is appealing to extend the circle representation to instance medical object segmentation. In this work, we propose CircleSnake, a simple end-to-end circle contour deformation-based segmentation method for ball-shaped medical objects. Compared to the prevalent DeepSnake method, our contribution is three-fold: (1) We replace the complicated bounding box to octagon contour transformation with a computation-free and consistent bounding circle to circle contour adaption for segmenting ball-shaped medical objects; (2) Circle representation has fewer degrees of freedom (DoF=2) as compared with the octagon representation (DoF=8), thus yielding a more robust segmentation performance and better rotation consistency; (3) To the best of our knowledge, the proposed CircleSnake method is the first end-to-end circle representation deep segmentation pipeline method with consistent circle detection, circle contour proposal, and circular convolution. The key innovation is to integrate the circular graph convolution with circle detection into an end-to-end instance segmentation framework, enabled by the proposed simple and consistent circle contour representation. Glomeruli are used to evaluate the performance of the benchmarks. From the results, CircleSnake increases the average precision of glomerular detection from 0.559 to 0.614. The Dice score increased from 0.804 to 0.849. The code has been released: https://github.com/hrlblab/CircleSnake
Large datasets that were made publicly available to the research community over the last 20 years have been a key enabling factor for the advances in deep learning algorithms for NLP or computer vision. These datasets are generally pairs of aligned image / manually annotated metadata, where images are photographs of everyday life. Scholarly and historical content, on the other hand, treat subjects that are not necessarily popular to a general audience, they may not always contain a large number of data points, and new data may be difficult or impossible to collect. Some exceptions do exist, for instance, scientific or health data, but this is not the case for cultural heritage (CH). The poor performance of the best models in computer vision - when tested over artworks - coupled with the lack of extensively annotated datasets for CH, and the fact that artwork images depict objects and actions not captured by photographs, indicate that a CH-specific dataset would be highly valuable for this community. We propose DEArt, at this point primarily an object detection and pose classification dataset meant to be a reference for paintings between the XIIth and the XVIIIth centuries. It contains more than 15000 images, about 80% non-iconic, aligned with manual annotations for the bounding boxes identifying all instances of 69 classes as well as 12 possible poses for boxes identifying human-like objects. Of these, more than 50 classes are CH-specific and thus do not appear in other datasets; these reflect imaginary beings, symbolic entities and other categories related to art. Additionally, existing datasets do not include pose annotations. Our results show that object detectors for the cultural heritage domain can achieve a level of precision comparable to state-of-art models for generic images via transfer learning.
Image Signal Processors (ISPs) play important roles in image recognition tasks as well as in the perceptual quality of captured images. In most cases, experts make a lot of effort to manually tune many parameters of ISPs, but the parameters are sub-optimal. In the literature, two types of techniques have been actively studied: a machine learning-based parameter tuning technique and a DNN-based ISP technique. The former is lightweight but lacks expressive power. The latter has expressive power, but the computational cost is too heavy on edge devices. To solve these problems, we propose "DynamicISP," which consists of multiple classical ISP functions and dynamically controls the parameters of each frame according to the recognition result of the previous frame. We show our method successfully controls the parameters of multiple ISP functions and achieves state-of-the-art accuracy with low computational cost in single and multi-category object detection tasks.
Despite monocular 3D object detection having recently made a significant leap forward thanks to the use of pre-trained depth estimators for pseudo-LiDAR recovery, such two-stage methods typically suffer from overfitting and are incapable of explicitly encapsulating the geometric relation between depth and object bounding box. To overcome this limitation, we instead propose OPA-3D, a single-stage, end-to-end, Occlusion-Aware Pixel-Wise Aggregation network that to jointly estimate dense scene depth with depth-bounding box residuals and object bounding boxes, allowing a two-stream detection of 3D objects, leading to significantly more robust detections. Thereby, the geometry stream denoted as the Geometry Stream, combines visible depth and depth-bounding box residuals to recover the object bounding box via explicit occlusion-aware optimization. In addition, a bounding box based geometry projection scheme is employed in an effort to enhance distance perception. The second stream, named as the Context Stream, directly regresses 3D object location and size. This novel two-stream representation further enables us to enforce cross-stream consistency terms which aligns the outputs of both streams, improving the overall performance. Extensive experiments on the public benchmark demonstrate that OPA-3D outperforms state-of-the-art methods on the main Car category, whilst keeping a real-time inference speed. We plan to release all codes and trained models soon.
Several SLAM methods benefit from the use of semantic information. Most integrate photometric methods with high-level semantics such as object detection and semantic segmentation. We propose that adding a semantic segmentation decoder in a shared encoder architecture would help the descriptor decoder learn semantic information, improving the feature extractor. This would be a more robust approach than only using high-level semantic information since it would be intrinsically learned in the descriptor and would not depend on the final quality of the semantic prediction. To add this information, we take advantage of multi-task learning methods to improve accuracy and balance the performance of each task. The proposed models are evaluated according to detection and matching metrics on the HPatches dataset. The results show that the Semantic SuperPoint model performs better than the baseline one.
Although modern object detectors rely heavily on a significant amount of training data, humans can easily detect novel objects using a few training examples. The mechanism of the human visual system is to interpret spatial relationships among various objects and this process enables us to exploit contextual information by considering the co-occurrence of objects. Thus, we propose a spatial reasoning framework that detects novel objects with only a few training examples in a context. We infer geometric relatedness between novel and base RoIs (Region-of-Interests) to enhance the feature representation of novel categories using an object detector well trained on base categories. We employ a graph convolutional network as the RoIs and their relatedness are defined as nodes and edges, respectively. Furthermore, we present spatial data augmentation to overcome the few-shot environment where all objects and bounding boxes in an image are resized randomly. Using the PASCAL VOC and MS COCO datasets, we demonstrate that the proposed method significantly outperforms the state-of-the-art methods and verify its efficacy through extensive ablation studies.
Among current anchor-based detectors, a positive anchor box will be intuitively assigned to the object that overlaps it the most. The assigned label to each anchor will directly determine the optimization direction of the corresponding prediction box, including the direction of box regression and category prediction. In our practice of crowded object detection, however, the results show that a positive anchor does not always regress toward the object that overlaps it the most when multiple objects overlap. We name it anchor drift. The anchor drift reflects that the anchor-object matching relation, which is determined by the degree of overlap between anchors and objects, is not always optimal. Conflicts between the fixed matching relation and learned experience in the past training process may cause ambiguous predictions and thus raise the false-positive rate. In this paper, a simple but efficient adaptive two-stage anchor assignment (TSAA) method is proposed. It utilizes the final prediction boxes rather than the fixed anchors to calculate the overlap degree with objects to determine which object to regress for each anchor. The participation of the prediction box makes the anchor-object assignment mechanism adaptive. Extensive experiments are conducted on three classic detectors RetinaNet, Faster-RCNN and YOLOv3 on CrowdHuman and COCO to evaluate the effectiveness of TSAA. The results show that TSAA can significantly improve the detectors' performance without additional computational costs or network structure changes.
We present a list of datasets and their best models with the goal of advancing the state-of-the-art in object detection by placing the question of object recognition in the context of the two types of state-of-the-art methods: one-stage methods and two stage-methods. We provided an in-depth statistical analysis of the five top datasets in the light of recent developments in granulated Deep Learning models - COCO minival, COCO test, Pascal VOC 2007, ADE20K, and ImageNet. The datasets are handpicked after closely comparing them with the rest in terms of diversity, quality of data, minimal bias, labeling quality etc. More importantly, our work extends to provide the best combination of these datasets with the emerging models in the last two years. It lists the top models and their optimal use cases for each of the respective datasets. We have provided a comprehensive overview of a variety of both generic and specific object detection models, enlisting comparative results like inference time and average precision of box (AP) fixed at different Intersection Over Union (IoUs) and for different sized objects. The qualitative and quantitative analysis will allow experts to achieve new performance records using the best combination of datasets and models.
In advanced paradigms of autonomous driving, learning Bird's Eye View (BEV) representation from surrounding views is crucial for multi-task framework. However, existing methods based on depth estimation or camera-driven attention are not stable to obtain transformation under noisy camera parameters, mainly with two challenges, accurate depth prediction and calibration. In this work, we present a completely Multi-Camera Calibration Free Transformer (CFT) for robust BEV representation, which focuses on exploring implicit mapping, not relied on camera intrinsics and extrinsics. To guide better feature learning from image views to BEV, CFT mines potential 3D information in BEV via our designed position-aware enhancement (PA). Instead of camera-driven point-wise or global transformation, for interaction within more effective region and lower computation cost, we propose a view-aware attention which also reduces redundant computation and promotes converge. CFT achieves 49.7% NDS on the nuScenes detection task leaderboard, which is the first work removing camera parameters, comparable to other geometry-guided methods. Without temporal input and other modal information, CFT achieves second highest performance with a smaller image input 1600 * 640. Thanks to view-attention variant, CFT reduces memory and transformer FLOPs for vanilla attention by about 12% and 60%, respectively, with improved NDS by 1.0%. Moreover, its natural robustness to noisy camera parameters makes CFT more competitive.
This work is for designing one-stage lightweight detectors which perform well in terms of mAP and latency. With baseline models each of which targets on GPU and CPU respectively, various operations are applied instead of the main operations in backbone networks of baseline models. In addition to experiments about backbone networks and operations, several feature pyramid network (FPN) architectures are investigated. Benchmarks and proposed detectors are analyzed in terms of the number of parameters, Gflops, GPU latency, CPU latency and mAP, on MS COCO dataset which is a benchmark dataset in object detection. This work propose similar or better network architectures considering the trade-off between accuracy and latency. For example, our proposed GPU-target backbone network outperforms that of YOLOX-tiny which is selected as the benchmark by 1.43x in speed and 0.5 mAP in accuracy on NVIDIA GeForce RTX 2080 Ti GPU.
Supervised object detection methods provide subpar performance when applied to Foreign Object Debris (FOD) detection because FOD could be arbitrary objects according to the Federal Aviation Administration (FAA) specification. Current supervised object detection algorithms require datasets that contain annotated examples of every to-be-detected object. While a large and expensive dataset could be developed to include common FOD examples, it is infeasible to collect all possible FOD examples in the dataset representation because of the open-ended nature of FOD. Limitations of the dataset could cause FOD detection systems driven by those supervised algorithms to miss certain FOD, which can become dangerous to airport operations. To this end, this paper presents a self-supervised FOD localization by learning to predict the runway images, which avoids the enumeration of FOD annotation examples. The localization method utilizes the Vision Transformer (ViT) to improve localization performance. The experiments show that the method successfully detects arbitrary FOD in real-world runway situations. The paper also provides an extension to the localization result to perform classification; a feature that can be useful to downstream tasks. To train the localization, this paper also presents a simple and realistic dataset creation framework that only collects clean runway images. The training and testing data for this method are collected at a local airport using unmanned aircraft systems (UAS). Additionally, the developed dataset is provided for public use and further studies.