Multiple Object Detection Research Articles

OptiRet-Net: An Optimized Low-Light Image Enhancement Technique for CV-based Applications in Resource-Constrained Environments

The illumination of images can significantly impact computer-vision applications such as image classification, multiple object detection, and tracking, leading to a significant decline in detection and tracking accuracy. Recent advancements in deep learning techniques have been applied to low-light image enhancement (LLIE) to combat this issue. Retinex theory-based methods following a decomposition-adjustment pipeline for LLIE have performed well in various aspects. Despite their success, current research on Retinex-based deep learning still needs to improve in terms of optimization techniques and complicated convolution connections, which can be computationally intensive for end-device deployment. We propose an Optimized Retinex-based CNN (OptiRet-Net) deep-learning framework to address these challenges for the low-light image enhancement problem. Our results demonstrate that the proposed method outperforms existing state-of-the-art models in terms of full reference metrics with a PSNR of 21.87, SSIM of 0.80, LPIPS of 0.16, and zero reference metrics with a NIQE of 3.4 and PIQE of 56.6. Additionally, we validate our approach using a comprehensive evaluation comprising five datasets and nine prior methods. Furthermore, we assess the efficacy of our proposed model combining low-light multiple object tracking applications using YOLOX and ByteTrack in versatile video coding (VVC/H.266) across various quantization parameters. Our findings reveal that LLIE-enhanced frames surpass their tracking results with a MOTA of 80.6% and a remarkable precision rate of 96%. Our model also achieves minimal file sizes by effectively compressing the enhanced low-light images while maintaining their quality, making it suitable for resource-constrained environments where storage or bandwidth limitations are a concern.

Automated particle inspection of continuously freeze-dried products using computer vision

The pharmaceutical industry is progressing towards more continuous manufacturing techniques. To dry biopharmaceuticals, continuous freeze drying has several advantages on manufacturing and process analytical control compared to batch freeze-drying, including better visual inspection potential. Visual inspection of every freeze-dried product is a key quality assessment after the lyophilization process to ensure that freeze-dried products are free from foreign particles and defects. This quality assessment is labor-intensive for operators who need to assess thousands of samples for an extensive amount of time leading to certain drawbacks. Applying Artificial Intelligence, specifically computer vision, on high-resolution images from every freeze-dried product can quantitatively and qualitatively outperform human visual inspection. For this study, continuously freeze-dried samples were prepared based on a real-world pharmaceutical product using manually induced particles of different sizes and subsequently imaged using a tailor-made setup to develop an image dataset (with particle sizes from 50μm to 1 mm) used to train multiple object detection models. You Only Look Once version 7 (YOLOv7) outperforms human inspection by a large margin, obtaining particle detection precision of up to 88.9% while controlling the recall at 81.2%, thus detecting most of the object present in the images, with an inference time of less than 1 s per vial.

A real-time and high-accuracy railway obstacle detection method using lightweight CNN and improved transformer

With the sustained development of railway transportation, the urgent necessity to improve train operation safety makes obstacle detection become the research focus. However, existing railway obstacle detectors still face challenges in balancing detection accuracy and speed during the shunting process. In addition, they are not robust enough in real-world railway environments, especially in complex scenes involving small obstacles. To address these problems, this paper presents a real-time and high-accuracy railway obstacle detection model using lightweight CNN and improved transformer (RH-Net) for detecting railway obstacles efficiently to guarantee traffic safety. First, the Lightweight Feature Extraction Module (LEM) is designed to minimize the model’s computational load while maintaining its feature extraction ability. Then, the Improved Transformer Module (IFM) is developed to boost the model’s ability about stably extract global contextual information. Finally, the Enhanced Multi-Scale Feature Fusion Module (EFM) is proposed to optimize the detection of obstacles with different sizes, especially small objects. In the experiments on railway dataset, RH-Net achieves optimal detection performance on GeForce GTX 1080Ti (96.99% mAP and 135 FPS) and Jetson Xavier NX (97.02% mAP and 43 FPS), which is significantly superior to the existing detection models. Experimental results show that RH-Net has excellent detection ability, which can accurately and efficiently detect obstacles in complicated railway environments. Moreover, the experiments on MS COCO indicate that RH-Net can achieve more satisfactory detection performance than existing state-of-the-art methods. Therefore, the proposed model can be well-applied to more complex real-world scenes for multiple object detection.

A new network model for multiple object detection for autonomous vehicle detection in mining environment

AbstractConsidering the challenges of low multi‐object detection accuracy and difficulty in identifying small targets caused by challenging environmental conditions including irregular lighting patterns and ambient noise levels in the mining environment with autonomous electric locomotives. A new network model based on SOD−YOLOv5s−4L has been proposed to detect multi‐objects for autonomous electric locomotives in underground coal mines. Improvements have been applied in YOLOv5s to construct the SOD−YOLOv5s−4L model, by introducing the SIoU loss function to address the mismatch between real and predicted bounding box directions, facilitating the model to learn target position information more efficiently. This research introduces a decoupled head to enhance feature fusion and improve the positioning precision of the network model, enabling rapid capture of multi‐scale target features. Furthermore, the detection capability of the model has been increased by introducing the small target detection layer which is developed by increasing the number of detection layers from three to four. The experimental results on multiple object detection dataset show that the proposed model achieves significant improvement in mean average precision (mAP) of almost 98% for various types of targets and an average precision (AP) of nearly 99% for small targets on the other hand it achieves 5.19% (mAP) and 9.79% (AP) compared to the YOLOv5s model. Furthermore, comparative analysis with other models like YOLOv7 and YOLOv8 shows that the proposed model has superior performance in terms of object detection.

Powering AI-driven car damage identification based on VeHIDE dataset

ABSTRACT In the realm of automobile insurance, the imperative of automating car damage evaluation has surged, offering streamlined assessment processes and heightened accuracy. Deep learning techniques have notably influenced vehicle damage assessment, reshaping insurance procedures. However, the primary challenge remains in crafting robust models for damage detection and segmentation. This study presents a novel contribution through the development of the Vehicle Damage Detection (VehiDE) dataset, specifically tailored for comprehensive car damage assessment. The dataset, encompassing 13,945 high-resolution images annotated across eight damage categories, serves as a foundational resource for advancing automated damage identification methodologies. Notably, VehiDE offers support for multiple tasks, including classification, object detection, instance segmentation, and salient object detection, thereby fostering versatile research avenues. Through extensive experimental analysis, including the evaluation of state-of-the-art methodologies on VehiDE, this study not only highlights the dataset's efficacy but also unveils new insights into the challenging nature of car damage assessment. Moreover, the study pioneers the exploration of salient object detection techniques in this domain, showcasing their potential in addressing irregular damage types. By offering VehiDE to the research community, we aim to catalyze advancements in the field of car damage assessment, paving the way for more accurate and efficient automated systems.

Decoupled Unbiased Teacher for Source-Free Domain Adaptive Medical Object Detection.

Source-free domain adaptation (SFDA) aims to adapt a lightweight pretrained source model to unlabeled new domains without the original labeled source data. Due to the privacy of patients and storage consumption concerns, SFDA is a more practical setting for building a generalized model in medical object detection. Existing methods usually apply the vanilla pseudo-labeling technique, while neglecting the bias issues in SFDA, leading to limited adaptation performance. To this end, we systematically analyze the biases in SFDA medical object detection by constructing a structural causal model (SCM) and propose an unbiased SFDA framework dubbed decoupled unbiased teacher (DUT). Based on the SCM, we derive that the confounding effect causes biases in the SFDA medical object detection task at the sample level, feature level, and prediction level. To prevent the model from emphasizing easy object patterns in the biased dataset, a dual invariance assessment (DIA) strategy is devised to generate counterfactual synthetics. The synthetics are based on unbiased invariant samples in both discrimination and semantic perspectives. To alleviate overfitting to domain-specific features in SFDA, we design a cross-domain feature intervention (CFI) module to explicitly deconfound the domain-specific prior with feature intervention and obtain unbiased features. Besides, we establish a correspondence supervision prioritization (CSP) strategy for addressing the prediction bias caused by coarse pseudo-labels by sample prioritizing and robust box supervision. Through extensive experiments on multiple SFDA medical object detection scenarios, DUT yields superior performance over previous state-of-the-art unsupervised domain adaptation (UDA) and SFDA counterparts, demonstrating the significance of addressing the bias issues in this challenging task. The code is available at https://github.com/CUHK-AIM-Group/Decoupled-Unbiased-Teacher.

Impact of Different Distance Metrics for Deep Learning-based multiple Object Detection and Tracking

Impact of Different Distance Metrics for Deep Learning-based multiple Object Detection and Tracking

Review: Multiple Objects Detection and Tracking using Deep Learning Approach

Multiple Object Tracking (MOT) is a crucial tool with diverse applications, such as object detection, object counting, and security systems. Precise identification and monitoring of numerous objects are essential in several computer vision uses, such as monitoring, self-driving cars, and computer-human communication. Very little has been done to address occlusion problems in order to enable the best moving object tracking with detection The tracking of visual objects is one of the most important components of computer vision. The process of tracking an object (or a group of objects) across time is called object tracking. Visual object tracking is used to identify or link target items over successive video frames. In this study, we analyze the tracking-by-detection strategy, which includes YOLO-based detection and SORT-based tracking. This work elucidates a general approach to tracking and recognizing many objects with an emphasis on accuracy improvement. We aim to revolutionize computer vision by applying Non-Maximum Suppression (NMS) and Intersection over Union (IoU) approaches, and by combining the state-of-the-art YOLO NAS algorithm with conventional tracking methods or an alternative version of the YOLO Algorithm for object identification. It is expected that our work will have a major impact on many different applications, enabling more precise and reliable object tracking and detection in difficult real-world scenarios. Keywords—Multiple Object Detection, Kalman Filters Multiple Object Tracking, DeepSORT, YOLO, IoU, NMS.

Remote intelligent perception system for multi-object detection.

During the last few years, a heightened interest has been shown in classifying scene images depicting diverse robotic environments. The surge in interest can be attributed to significant improvements in visual sensor technology, which has enhanced image analysis capabilities. Advances in vision technology have a major impact on the areas of multiple object detection and scene understanding. These tasks are an integral part of a variety of technologies, including integrating scenes in augmented reality, facilitating robot navigation, enabling autonomous driving systems, and improving applications in tourist information. Despite significant strides in visual interpretation, numerous challenges persist, encompassing semantic understanding, occlusion, orientation, insufficient availability of labeled data, uneven illumination including shadows and lighting, variation in direction, and object size and changing background. To overcome these challenges, we proposed an innovative scene recognition framework, which proved to be highly effective and yielded remarkable results. First, we perform preprocessing using kernel convolution on scene data. Second, we perform semantic segmentation using UNet segmentation. Then, we extract features from these segmented data using discrete wavelet transform (DWT), Sobel and Laplacian, and textual (local binary pattern analysis). To recognize the object, we have used deep belief network and then find the object-to-object relation. Finally, AlexNet is used to assign the relevant labels to the scene based on recognized objects in the image. The performance of the proposed system was validated using three standard datasets: PASCALVOC-12, Cityscapes, and Caltech 101. The accuracy attained on the PASCALVOC-12 dataset exceeds 96% while achieving a rate of 95.90% on the Cityscapes dataset. Furthermore, the model demonstrates a commendable accuracy of 92.2% on the Caltech 101 dataset. This model showcases noteworthy advancements beyond the capabilities of current models.

CA-YOLO: Ensembled Model Optimization for Remote Sensing Image Object Detection

Multiple object retrieval methods are solved by the CA-YOLO model to find elements in complex spatial images. This addresses weak multiscale feature learning and the delicate balance between model complexity and performance. CA-YOLO is based on YOLOv5 and uses a small coordinate attention module at the lower layer to recover selected features and reduce data. A fast spatial pyramid with tandem design modules at the bottom uses stochastic pooling to speed up thinking and merge features of different sizes. Anchor box interactions and missing features have been updated to make it easier to find items of different sizes and shapes. CA-YOLO outperforms YOLO in multiple object detection and average inference speed of 125 frames/sec. CA-YOLO is a great option in the same conditions and difficulties. The study also investigated V3 tiny, V4, V5s, V8s, CA Yolos, and V5x6 YOLO models. This suggests that YOLO V5x6 can achieve more than 95% mAP on remote sensing object identification datasets. Index terms - Object detection, attention mechanism, coordinate attention, SPPF, SIoU loss.

Detection of Crabs and Lobsters Using a Benchmark Single-Stage Detector and Novel Fisheries Dataset

Crabs and lobsters are valuable crustaceans that contribute enormously to the seafood needs of the growing human population. This paper presents a comprehensive analysis of single- and multi-stage object detectors for the detection of crabs and lobsters using images captured onboard fishing boats. We investigate the speed and accuracy of multiple object detection techniques using a novel dataset, multiple backbone networks, various input sizes, and fine-tuned parameters. We extend our work to train lightweight models to accommodate the fishing boats equipped with low-power hardware systems. Firstly, we train Faster R-CNN, SSD, and YOLO with different backbones and tuning parameters. The models trained with higher input sizes resulted in lower frames per second (FPS) and vice versa. The base models were highly accurate but were compromised in computational and run-time costs. The lightweight models were adaptable to low-power hardware compared to the base models. Secondly, we improved the performance of YOLO (v3, v4, and tiny versions) using custom anchors generated by the k-means clustering approach using our novel dataset. The YOLO (v4 and it’s tiny version) achieved mean average precision (mAP) of 99.2% and 95.2%, respectively. The YOLOv4-tiny trained on the custom anchor-based dataset is capable of precisely detecting crabs and lobsters onboard fishing boats at 64 frames per second (FPS) on an NVidia GeForce RTX 3070 GPU. The Results obtained identified the strengths and weaknesses of each method towards a trade-off between speed and accuracy for detecting objects in input images.

Analytical Formalism for Data Representation and Object Detection with 2D LiDAR: Application in Mobile Robotics.

In mobile robotics, LASER scanners have a wide spectrum of indoor and outdoor applications, both in structured and unstructured environments, due to their accuracy and precision. Most works that use this sensor have their own data representation and their own case-specific modeling strategies, and no common formalism is adopted. To address this issue, this manuscript presents an analytical approach for the identification and localization of objects using 2D LiDARs. Our main contribution lies in formally defining LASER sensor measurements and their representation, the identification of objects, their main properties, and their location in a scene. We validate our proposal with experiments in generic semi-structured environments common in autonomous navigation, and we demonstrate its feasibility in multiple object detection and identification, strictly following its analytical representation. Finally, our proposal further encourages and facilitates the design, modeling, and implementation of other applications that use LASER scanners as a distance sensor.

Uncertainty-driven active developmental learning

Existing machine learning models can well handle common classes but struggle to detect unfamiliar or unknown classes due to environmental variations. To address this challenge, we propose a new task called active developmental learning (ADL), which empowers models to actively determine what to learn in the open world, thereby progressively enhancing the capability of detecting unfamiliar and unknown classes. Considering the uncertain essence of the task, we design an uncertainty-driven method for ADL (UADL) that measures and utilizes uncertainty to evaluate unfamiliar known classes and unknown classes separately, which consists of two stages: (1) unfamiliar detection of known classes and (2) unknown detection of novel classes. In the first stage, UADL identifies unfamiliar samples of known classes via known-class uncertainty calculated by GMMs on detectors’ heads. In the second stage, UADL identifies samples containing unknown classes via unknown-class uncertainty computed by class-specific GMMs in feature space. In both stages, uncertainty is used to select a minimal number of unlabeled samples for manual labeling, facilitating the model’s active self-development. Experiments on multiple object detection benchmark datasets demonstrate the feasibility and significant performance of UADL and show its effectiveness against the ADL task compared to other state-of-the-art approaches.

Multiple object detection and tracking from drone videos based on GM-YOLO and multi-tracker

Multiple object tracking in drone videos is a vital vision task with broad application prospects, but most trackers use spatial or appearance clues alone to correlate detections. Our proposed Multi-Tracker uses a novel similarity measure that combines position and appearance information. We designed the GM-YOLO network to provide high-quality detections as input to Multi-Tracker. Add a Coordinate Attention mechanism and a weighted Bidirectional Feature Pyramid Network structure to the Backbone, each feature point's effective receptive field is modeled as a Gaussian distribution. To accurately obtain the motion and appearance features of the object, the adaptive noise covariance Kalman filter is used to get the position information, MB-OSNet network is designed to use global features to learn contour information to retrieve images from a wider field of view while incorporating Part-Level elements that contain more fine-grained data. Finally, the motion and appearance features are jointly compared to realize multi object tracking. The performance of the GM-YOLO object detector and the Multi-Tracker was verified on the VisDrone MOT and UAVDT datasets.

RoboCoV Cleaner: An Indoor Autonomous UV-C Disinfection Robot with Advanced Dual-Safety Systems.

In the face of today's ever-evolving global health landscape and ambient assisted living (AAL), marked by the persistent emergence of novel viruses and diseases that impact vulnerable categories and individual safety, the need for innovative disinfection solutions has surged to unprecedented levels. In pursuit of advancing the field of autonomous UV-C disinfection robotics, we conducted two comprehensive state-of-the-art analyses: the first one in the literature and the second one in existing commercial disinfection robots to identify current challenges. Of all of the challenges, we consider the most outstanding ones to be safeguarding humans and animals and understanding the surroundings while operating the disinfection process challenges that we will address in this article. While UV-C lamps have demonstrated their effectiveness in sterilizing air and surfaces, the field of autonomous UV-C disinfection robotics represents a critical domain that requires advancement, particularly in safeguarding the wellbeing of humans and animals during operation. Operating UV-C disinfection robots in close proximity to humans or animals introduces inherent risks, and existing disinfection robots often fall short in incorporating advanced safety systems. In response to these challenges, we propose the RoboCoV Cleaner-an indoor autonomous UV-C disinfection robot equipped with an advanced dual and redundant safety system. This novel approach incorporates multiple passive infrared (PIR) sensors and AI object detection on a 360-degree camera. Under our test, the dual-redundant system reached more than 90% when detecting humans with high accuracy using the AI system 99% up to 30 m away in a university hallway (different light conditions) combined with the PIR system (with lower accuracy). The PIR system was proved to be a redundant system for uninterrupted operation during communication challenges, ensuring continuous sensor information collection with a swift response time of 50 ms (image processing within 200 ms). It empowers the robot to detect and react to human presence, even under challenging conditions, such as when individuals wear masks, in complete darkness, under UV light, or in environments with blurred visual conditions. In our test, the detection system performed outstandingly well with up to 99% detection rate of humans. Beyond safety features, the RoboCoV Cleaner can identify objects in its surroundings. This capability empowers the robot to discern objects affected by UV-C light, enabling it to apply specialized rules for targeted disinfection. The proposed system exhibits a wide range of capabilities beyond its core purpose of disinfection, making it suitable for healthcare facilities, universities, conference venues, and hospitals. Its implementation has the ability to improve significantly human safety and protect people. By showcasing the RoboCoV Cleaner's safety-first approach and adaptability, we aim to set a new benchmark for UV-C disinfection robots, promoting clean and secure environments while protecting vulnerable people, even in challenging scenarios.

Image Filtering Techniques for Object Recognition in Autonomous Vehicles

The deployment of autonomous vehicles has the potential to significantly lessen the variety of current harmful externalities, (such as accidents, traffic congestion, security, and environmental degradation), making autonomous vehicles an emerging topic of research. In this paper, a literature review of autonomous vehicle development has been conducted with a notable finding that autonomous vehicles will inevitably become an indispensable future greener solution. Subsequently, 5 different deep learning models, YOLOv5s, EfficientNet-B7, Xception, MobilenetV3, and InceptionV4, have been built and analyzed for 2-D object recognition in the navigation system. While testing on the BDD100K dataset, YOLOv5s and EfficientNet-B7 appear to be the two best models. Finally, this study has proposed Hessian, Laplacian, and Hessian-based Ridge Detection filtering techniques to optimize the performance of those 2 models. The results demonstrate that these filters could increase the mean average precision by up to 11.81%, and reduce detection time by up to 43.98% when applied to YOLOv5s and EfficientNet-B7 models. Overall, all the experiment results are promising and could be extended to other domains for semantic understanding of the environment. Additionally, various filtering algorithms for multiple object detection and classification could be applied to other areas. Different recommendations and future work have been clearly defined in this study.

Solving traffic data occlusion problems in computer vision algorithms using DeepSORT and quantum computing

Inaccuracies of traffic sensors during traffic counting and vehicle classification have persisted as transportation agencies have been prompted to calibrate sensors periodically. Detection of multiple objects, heavy occlusions, and similar appearances in congested places are some causes of computer vision model inaccuracies. This paper used the YOLOv5 model for detection and the DeepSORT model for tracking objects. Due to the nature of the reported problem caused by many misses and mismatches, the power of quantum computing with the alternating direction method of multipliers (ADMM) optimizer was leveraged. A basic Kalman filter and the Hungarian algorithm features were used in combination with a quantum optimizer to present robust multiple object tracking (MOT) algorithms. This hybrid combination of the classical and quantum model has fastened learning the occludes during frame matching of tracks and detections by generating minimum quantum cost function value. Comparisons with the existing models indicated a significant increase in the primary MOT metric multiple object tracking accuracy (MOTA) by 16% more than the regular YOLOv5-DeepSORT model when using a quantum optimizer. Also, a 6% multiple object tracking precision (MOTP) increases and a 6% identification metrics (F1) score increase were observed using the quantum optimizer with identity switching reduced from 6 to 4. This model is expected to assist transportation officials in improving the accuracy of traffic counts and vehicle classification and reduce the need for regular computer vision software calibration.

Individual honey bee tracking in a beehive environment using deep learning and Kalman filter

The honey bee is the most essential pollinator and a key contributor to the natural ecosystem. There are numerous ways for thousands of bees in a hive to communicate with one another. Individual trajectories and social interactions are thus complex behavioral features that can provide valuable information for an ecological study. To study honey bee behavior, the key challenges that have resulted from unreliable studies include complexity (high density of similar objects, small objects, and occlusion), the variety of background scenes, the dynamism of individual bee movements, and the similarity between the bee body and the background in the beehive. This study investigated the tracking of individual bees in a beehive environment using a deep learning approach and a Kalman filter. Detection of multiple bees and individual object segmentation were performed using Mask R-CNN with a ResNet-101 backbone network. Subsequently, the Kalman filter was employed for tracking multiple bees by tracking the body of each bee across a sequence of image frames. Three metrics were used to assess the proposed framework: mean average precision (mAP) for multiple-object detection and segmentation tasks, CLEAR MOT for multiple object tracking tasks, and MOTS for multiple object tracking and segmentation tasks. For CLEAR MOT and MOTS metrics, accuracy (MOTA and MOTSA) and precision (MOTP and MOTSP) are considered. By employing videos from a custom-designed observation beehive, recorded at a frame rate of 30 frames per second (fps) and utilizing a continuous frame rate of 10 fps as input data, our system displayed impressive performance. It yielded satisfactory outcomes for tasks involving segmentation and tracking of multiple instances of bee behavior. For the multiple-object segmentation task based on Mask R-CNN, we achieved a 0.85 mAP. For the multiple-object-tracking task with the Kalman filter, we achieved 77.48% MOTA, 79.79% MOTSP, and 79.56% recall. For the overall system for multiple-object tracking and segmentation tasks, we achieved 77.00% MOTSA, 75.60% MOTSP, and 80.30% recall.

Оптимизация обнаружения удаленных объектов с помощью моностатических радарных устройств оптического диапазона

The article discusses optimization of detection of multiple remote objects in the infrared range by means of monostatic optical radars. Optimization problems are formulated and solved that allow increasing the efficiency of detection of multiple remote objects in the optical range. The conditions for achieving the minimum average value of the cross section of detectable objects are determined in two modes: providing the inverse dependence of the detection distance on the minimum value of the detected signal from the targets; providing the direct dependence of the detection distance on the wavelength, i. e. in those areas where the atmosphere transmission increases with increasing wavelength.

OffsetNet: Towards Efficient Multiple Object Tracking, Detection, and Segmentation.

Offset-based representation has emerged as a promising approach for modeling semantic relations between pixels and object motion, demonstrating efficacy across various computer vision tasks. In this paper, we introduce a novel one-stage multi-tasking network tailored to extend the offset-based approach to MOTS. Our proposed framework, named OffsetNet, is designed to concurrently address amodal bounding box detection, instance segmentation, and tracking. It achieves this by formulating these three tasks within a unified pixel-offset-based representation, thereby achieving excellent efficiency and encouraging mutual collaborations. OffsetNet achieves several remarkable properties: first, the encoder is empowered by a novel Memory Enhanced Linear Self-Attention (MELSA) block to efficiently aggregate spatial-temporal features; second, all tasks are decoupled fairly using three lightweight decoders that operate in a one-shot manner; third, a novel cross-frame offsets prediction module is proposed to enhance the robustness of tracking against occlusions. With these merits, OffsetNet achieves 76.83% HOTA on KITTI MOTS benchmark, which is the best result without relying on 3D detection. Furthermore, OffsetNet achieves 74.83% HOTA at 50 FPS on the KITTI MOT benchmark, which is nearly 3.3 times faster than CenterTrack with better performance. We hope our approach will serve as a solid baseline and encourage future research in this field.