Bounding Box Annotations Research Articles

WeakPCSOD: Overcoming the Bias of Box Annotations for Weakly Supervised Point Cloud Salient Object Detection

Point cloud salient object detection (PCSOD) is a newly proposed task in 3D dense segmentation. However, the acquisition of accurate 3D dense annotations comes at a high cost, severely limiting the progress of PCSOD. To address this issue, we propose the first weakly supervised PCSOD (named WeakPCSOD) model, which relies solely on cheap 3D bounding box annotations. In WeakPCSOD, we extract noise-free supervision from coarse 3D bounding boxes while mitigating shape biases inherent in box annotations. To achieve this, we introduce a novel mask-to-box (M2B) transformation and a color consistency (CC) loss. The M2B transformation, from a shape perspective, disentangles predictions from labels, enabling the extraction of noiseless supervision from labels while preserving object shapes independently of the box bias. From an appearance perspective, we further introduce the CC loss to provide dense supervision, which mitigates the non-unique predictions stemming from weak supervision and substantially reduces prediction variability. Furthermore, we employ a self-training (ST) strategy to enhance performance by utilizing high-confidence pseudo labels. Notably, the M2B transformation, CC loss, and ST strategy are seamlessly integrated into any model and incur no computational costs for inference. Extensive experiments demonstrate the effectiveness of our WeakPCSOD model, even comparable to fully supervised models utilizing dense annotations.

Detection of Ulcerative Colitis Lesions from Weakly Annotated Colonoscopy Videos Using Bounding Boxes

Ulcerative colitis is a chronic disease characterized by bleeding and ulcers in the colon. Disease severity assessment via colonoscopy videos is time-consuming and only focuses on the most severe lesions. Automated detection methods enable fine-grained assessment but depend on the training set quality. To suit the local clinical setup, an internal training dataset containing only rough bounding box annotations around lesions was utilized. Following previous works, we propose to use linear models in suitable color spaces to detect lesions. We introduce an efficient sampling scheme for exploring the set of linear classifiers and removing trivial models i.e., those showing zero false negative or positive ratios. Bounding boxes lead to exaggerated false detection ratios due to mislabeled pixels, especially in the corners, resulting in decreased model accuracy. Therefore, we propose to evaluate the model sensitivity on the annotation level instead of the pixel level. Our sampling strategy can eliminate up to 25% of trivial models. Despite the limited quality of annotations, the detectors achieved better performance in comparison with the state-of-the-art methods. When tested on a small subset of endoscopic images, the best models exhibit low variability. However, the inter-patient model performance was variable suggesting that appearance normalization is critical in this context.

Learning From Box Annotations for Referring Image Segmentation.

Referring image segmentation (RIS) has obtained an impressive achievement by fully convolutional networks (FCNs). However, previous RIS methods require a large number of pixel-level annotations. In this article, we present a weakly supervised RIS method by using bounding box (BB) annotations. In the first stage, we introduce an adversarial boundary loss to extract the object contour from the BB, which is then used to select appropriate region proposals for pseudoground-truth (PGT) generation. In the second stage, we design a co-training (Co-T) strategy to purify the pseudolabels. Specifically, we train two networks and interactively guide them to pick clean labels for each other's networks, which can weaken the effect of noisy labels on model training. Experiment results on four benchmark datasets demonstrate that the proposed method can produce high-quality masks with a speed of 63 frames/s.

OBBInst: Remote sensing instance segmentation with oriented bounding box supervision

Remote sensing (RS) instance segmentation is an important but challenging task due to multi-oriented, densely arranged objects and lack of mask annotation. Compared with redundant horizontal bounding-box (HBB) and expensive pixel-level annotation, oriented bounding box (OBB) annotations can provide compact object depicts with lower annotation costs. Therefore, we propose the first weakly supervised remote sensing instance segmentation method with OBB supervision (namely OBBInst) to reduce the annotation burden and make full use of existing abundant OBB annotations. Based on BoxInst (a high-performance instance segmentation method with box annotations), OBBInst has customized a framework for OBB annotation to unify the incompatibility between existing HBB-based and OBB-based methods. In addition, we propose an oriented projection method with a corresponding loss function to achieve more precise target depicts of OBB annotation. Moreover, we propose an edge similarity loss to incorporate Canny edge prior into deep learning framework for more precise edge identification of densely arranged objects. We have conducted extensive experiments on iSAID and HRSC datasets, and the experimental results demonstrate that OBBInst can achieve the state-of-the-art performance as compared to existing box-supervised methods. In addition, OBBInst dramatically narrows the performance gap between weakly and fully supervised instance segmentation (23.9% vs. 35.6% in iSAID dataset and 79.5% vs. 84.9% in HRSC dataset).

WaterPairs: a paired dataset for underwater image enhancement and underwater object detection

Due to its importance in marine engineering and aquatic robotics, underwater image enhancement works as a preprocessing step to improve the performance of high-level vision tasks such as underwater object detection and recognition. Although several studies exhibit that underwater image enhancement algorithms can boost the detection accuracy of detectors, no work has focused on studying the relationship between these two tasks. This is mainly because current underwater datasets lack either bounding box annotations or high-quality reference images, based on which detection accuracy or image quality assessment metrics are calculated. To examine how underwater image enhancement methods affect underwater object detection tasks, we provide a large-scale underwater object detection dataset with both bounding box annotations and high-quality reference images, namely, the WaterPairs dataset. The WaterPairs dataset offers a platform for researchers to comprehensively study the influence of underwater image enhancement algorithms on underwater object detection tasks. We will release our dataset at https://github.com/IanDragon/WaterPairs once this paper is accepted.

AriAplBud: An Aerial Multi-Growth Stage Apple Flower Bud Dataset for Agricultural Object Detection Benchmarking

As one of the most important topics in contemporary computer vision research, object detection has received wide attention from the precision agriculture community for diverse applications. While state-of-the-art object detection frameworks are usually evaluated against large-scale public datasets containing mostly non-agricultural objects, a specialized dataset that reflects unique properties of plants would aid researchers in investigating the utility of newly developed object detectors within agricultural contexts. This article presents AriAplBud: a close-up apple flower bud image dataset created using an unmanned aerial vehicle (UAV)-based red–green–blue (RGB) camera. AriAplBud contains 3600 images of apple flower buds at six growth stages, with 110,467 manual bounding box annotations as positive samples and 2520 additional empty orchard images containing no apple flower bud as negative samples. AriAplBud can be directly deployed for developing object detection models that accept Darknet annotation format without additional preprocessing steps, serving as a potential benchmark for future agricultural object detection research. A demonstration of developing YOLOv8-based apple flower bud detectors is also presented in this article.

An Underwater Organism Image Dataset and a Lightweight Module Designed for Object Detection Networks

Long-term monitoring and recognition of underwater organism objects are of great significance in marine ecology, fisheries science and many other disciplines. Traditional techniques in this field, including manual fishing-based ones and sonar-based ones, are usually flawed. Specifically, the method based on manual fishing is time-consuming and unsuitable for scientific researches, while the sonar-based one, has the defects of low acoustic image accuracy and large echo errors. In recent years, the rapid development of deep learning and its excellent performance in computer vision tasks make vision-based solutions feasible. However, the researches in this area are still relatively insufficient in mainly two aspects. First, to our knowledge, there is still a lack of large-scale datasets of underwater organism images with accurate annotations. Second, in consideration of the limitation on hardware resources of underwater devices, an underwater organism detection algorithm that is both accurate and lightweight enough to be able to infer in real time is still lacking. As an attempt to fill in the aforementioned research gaps to some extent, we established the Multiple Kinds of Underwater Organisms (MKUO) dataset with accurate bounding box annotations of taxonomic information, which consists of 10,043 annotated images, covering eighty-four underwater organism categories. Based on our benchmark dataset, we evaluated a series of existing object detection algorithms to obtain their accuracy and complexity indicators as the baseline for future reference. In addition, we also propose a novel lightweight module, namely Sparse Ghost Module, designed especially for object detection networks. By substituting the standard convolution with our proposed one, the network complexity can be significantly reduced and the inference speed can be greatly improved without obvious detection accuracy loss. To make our results reproducible, the dataset and the source code are available online at https://cslinzhang.github.io/MKUO-and-Sparse-Ghost-Module/ .

Unsupervised Logo Detection Using Adversarial Learning From Synthetic to Real Images

Most existing deep learning-based methods for logo detection require a large amount of object-level annotated training data. However, it takes a lot of time and effort to create a good object-level annotated logo detection dataset. Many researchers have tried to utilize synthesized images with automatically generated annotations for training to deal with data annotation problem. However, in real applications, model trained using synthesized images suffer from performance loss due to domain shift between synthesized and real-world images. To this end, in this paper, we propose an adversarial learning-based unsupervised domain adaptation method for logo detection. We only use labeled synthesized logo images for model training and adapt target domain knowledge using unlabeled real-world logo images. We propose entropy minimization of mid-level output feature maps in order to effectively align the domain gap between synthetic and real images for the logo detection task. Additionally, we have generated coherent synthesized logo images with automatically constructed bounding box annotations for different datasets to perform unsupervised training for the experiments. Our experiments show that the proposed method improves performance on different logo datasets compared to direct transfer from source to target domain (synthetic-to-real images) without any labeling cost and increasing network parameters.

PanorAMS: Automatic Annotation for Detecting Objects in Urban Context

Large collections of geo-referenced panoramic images are freely available for cities across the globe, as well as detailed maps with location and meta-data on a great variety of urban objects. They provide a potentially rich source of information on urban objects, but manual annotation for object detection is expensive, laborious and challenging. Can we utilize such multimedia sources to automatically annotate street level images as an inexpensive alternative to manual labeling? With the PanorAMS framework we introduce a method to automatically generate bounding box annotations for panoramic images based on urban context information. Following this method, we acquire large-scale, albeit noisy, annotations for an urban dataset solely from open data sources in a fast and automatic manner. The dataset covers the City of Amsterdam and includes over 14 million noisy bounding box annotations of 22 object categories present in 771,299 panoramic images. For many objects further fine-grained information is available, obtained from geospatial meta-data, such as <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">building value</i> , <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">function</i> and <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">average surface area</i> . Such information would have been difficult, if not impossible, to acquire via manual labeling based on the image alone. For detailed evaluation, we introduce an efficient crowdsourcing protocol for bounding box annotations in panoramic images, which we deploy to acquire 147,075 ground-truth object annotations for a subset of 7,348 images, the PanorAMS-clean dataset. For our PanorAMS-noisy dataset, we provide an extensive analysis of the noise and how different types of noise affect image classification and object detection performance. We make both datasets, PanorAMS-noisy and PanorAMS-clean, benchmarks and tools presented in this paper openly available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://panorams.inskegroenen.nl/</uri> .

Weakly supervised salient object detection via bounding-box annotation and SAM model

&lt;abstract&gt;&lt;p&gt;Salient object detection (SOD) aims to detect the most attractive region in an image. Fully supervised SOD based on deep learning usually needs a large amount of data with human annotation. Researchers have gradually focused on the SOD task using weakly supervised annotation such as category, scribble, and bounding-box, while these existing weakly supervised methods achieve limited performance and demonstrate a huge performance gap with fully supervised methods. In this work, we proposed one novel two-stage weakly supervised method based on bounding-box annotation and the recent large visual model Segment Anything (SAM). In the first stage, we regarded the bounding-box annotation as the box prompt of SAM to generate initial labels and proposed object completeness check and object inversion check to exclude low quality labels, then we selected reliable pseudo labels for the training initial SOD model. In the second stage, we used the initial SOD model to predict the saliency map of excluded images and adopted SAM with the everything mode to generate segmentation candidates, then we fused the saliency map and segmentation candidates to predict pseudo labels. Finally we used all reliable pseudo labels generated in the two stages to train one refined SOD model. We also designed a simple but effective SOD model, which can capture rich global context information. Performance evaluation on four public datasets showed that the proposed method significantly outperforms other weakly supervised methods and also achieves comparable performance with fully supervised methods.&lt;/p&gt;&lt;/abstract&gt;

Progressive Frame-Proposal Mining for Weakly Supervised Video Object Detection.

In this paper, we focus on the weakly supervised video object detection problem, where each training video is only tagged with object labels, without any bounding box annotations of objects. To effectively train object detectors from such weakly-annotated videos, we propose a Progressive Frame-Proposal Mining (PFPM) framework by exploiting discriminative proposals in a coarse-to-fine manner. First, we design a flexible Multi-Level Selection (MLS) scheme, with explicit guidance of video tags. By selecting object-relevant frames and mining important proposals from these frames, the proposed MLS can effectively reduce frame redundancy as well as improve proposal effectiveness to boost weakly-supervised detectors. Moreover, we develop a novel Holistic-View Refinement (HVR) scheme, which can globally evaluate importance of proposals among frames, and thus correctly refine pseudo ground truth boxes for training video detectors in a self-supervised manner. Finally, we evaluate the proposed PFPM on a large-scale benchmark for video object detection, on ImageNet VID, under the setting of weak annotations. The experimental results demonstrate that our PFPM significantly outperforms the state-of-the-art weakly-supervised detectors.

Go Deep or Broad? Exploit Hybrid Network Architecture for Weakly Supervised Object Classification and Localization.

Weakly supervised object classification and localization are learned object classes and locations using only image-level labels, as opposed to bounding box annotations. Conventional deep convolutional neural network (CNN)-based methods activate the most discriminate part of an object in feature maps and then attempt to expand feature activation to the whole object, which leads to deteriorating the classification performance. In addition, those methods only use the most semantic information in the last feature map, while ignoring the role of shallow features. So, it remains a challenge to enhance classification and localization performance with a single frame. In this article, we propose a novel hybrid network, namely deep and broad hybrid network (DB-HybridNet), which combines deep CNNs with a broad learning network to learn discriminative and complementary features from different layers, and then integrates multilevel features (i.e., high-level semantic features and low-level edge features) in a global feature augmentation module. Importantly, we exploit different combinations of deep features and broad learning layers in DB-HybridNet and design an iterative training algorithm based on gradient descent to ensure the hybrid network work in an end-to-end framework. Through extensive experiments on caltech-UCSD birds (CUB)-200 and imagenet large scale visual recognition challenge (ILSVRC) 2016 datasets, we achieve state-of-the-art classification and localization performance.

An Effective Dataset Preprocessing Method in Tilted Gear Defects Target Detection

Gear defect detection is a crucial component in power automation systems. Methods based on deep learning have exhibited excellent performance in detecting gears. However, the effect of defect detection for tilted gears is not as good. This is due to the traditional horizontal bounding box annotation method, which inevitably produces many overlapping areas in the annotated bounding box when used in tilted gear targets, and a large part of the areas do not belong to defects. In order to address the issue of low precision in the detection of defects in tilted gears based on deep learning, a dataset preprocessing scheme has been proposed. Initially, the images and annotation files of the training set and validation set are automatically rotated. Subsequently, the rotated data are utilized to train the defect detection model. Finally, the test set is subjected to the same image rotation method as the first step, resulting in the generation of high‐precision defect detection results, which are then input into the defect detection model. In order to facilitate comparison with the ground truth, the JSON file obtained from the detection result is rotated in reverse and the result is mapped to the original test set before rotation. In order to verify the effectiveness of the dataset optimization method proposed in this article, the same configuration file is used to train and evaluate the gear defect detection model with the dataset before and after optimization, respectively. The three detection models with the highest comprehensive indicators were employed to assess the dataset before and after optimization, with the average detection precision mAP serving as a quantitative comparison indicator. The findings demonstrate that the data optimization method proposed in this article has markedly enhanced mAP in tilted gear defect detection. Upon testing the proposed method on the PP‐YOLOE + model, mAP (0.5) exhibited an increase of 18.5%, while mAP (0.5 : 0.95) demonstrated a 29.1% improvement. When the method was tested on the RT‐DETR model, mAP (0.5) exhibited an increase of 24.7%, while mAP (0.5 : 0.95) demonstrated an 18.4% improvement. When tested on the few‐short model, mAP (0.5) increased by 23.3% and mAP (0.5 : 0.95) increased by 19.8%. This validates the effectiveness of the dataset processing method proposed in this article in the detection of tilted gear defects.

UAS and UGV-Based Disease Management System for Diagnosing Corn Diseases Above and Below the Canopy Using Deep Learning

Highlights Generated custom imagery dataset using UAS, UGV, and handheld sensors separately in diseased corn fields. Proposed data pipeline utilizing Google Sheets API to establish communication between each platform and enable access through a web application. Developed deep learning-based disease management system for above and below-the-canopy corn disease diagnosis. Trained and evaluated disease detection models from each platform separately to provide management recommendations. Abstract. Early disease management following the onset of disease symptoms is crucial for controlling their spread. Heterogenous collaboration between unmanned aerial systems (UAS) and unmanned ground vehicles (UGV) for field scouting and disease diagnosis is a potential approach for developing automated disease management solutions. However, automation of crop-specific disease identification requires the use of above and below-canopy sensors and properly trained deep learning (DL) models. This research proposes to develop a novel disease management system for diagnosing corn diseases from above and below the canopy by collaboratively using edge devices mounted on UAS and UGV, respectively. Three separate datasets were acquired using UAS above the canopy, UGV below the canopy, and handheld imaging platforms within diseased corn fields. DL-based image classification models were first trained for identifying common corn diseases under field conditions, resulting in up to 95.04% testing accuracy using the DenseNet169 architecture. After creating bounding box annotations for disease images, You Only Look Once (YOLO)v7 DL-based object detection models were trained to identify diseases from each platform separately. Trained YOLOv7 models resulted in the highest mAP@IoU=0.5 of 37.6%, 46.4%, and 72.2% for locating and identifying diseases above the canopy using UAS, below the canopy using UGV, and handheld sensors, respectively. A client/server architecture was developed to establish communication between the UAS, UGV, and Google Spreadsheets via Wi-Fi communication protocol. The coordinates of diseased regions and distinct disease types were recorded using the client/server architecture on Google Spreadsheets. A web application utilized the data from the Google Spreadsheet to help users diagnose diseases in real time and provide recommendations for implementing appropriate disease management practices. This study reports findings of independently operated UAS and UGV that can potentially offer disease spread from below and over the corn canopy by combining the information. Keywords: Deep learning, Disease identification, Disease management, Object detection, UAS, UGV, YOLOv7.

Weakly supervised image segmentation beyond tight bounding box annotations

Weakly supervised image segmentation approaches in the literature usually achieve high segmentation performance using tight bounding box supervision and decrease the performance greatly when supervised by loose bounding boxes. However, compared with loose bounding box, it is much more difficult to acquire tight bounding box due to its strict requirements on the precise locations of the four sides of the box. To resolve this issue, this study investigates whether it is possible to maintain good segmentation performance when loose bounding boxes are used as supervision. For this purpose, this work extends our previous parallel transformation based multiple instance learning (MIL) for tight bounding box supervision by integrating an MIL strategy based on polar transformation to assist image segmentation. The proposed polar transformation based MIL formulation works for both tight and loose bounding boxes, in which a positive bag is defined as pixels in a polar line of a bounding box with one endpoint located inside the object enclosed by the box and the other endpoint located at one of the four sides of the box. Moreover, a weighted smooth maximum approximation is introduced to incorporate the observation that pixels closer to the origin of the polar transformation are more likely to belong to the object in the box. The proposed approach was evaluated on two public datasets using dice coefficient when bounding boxes at different precision levels were considered in the experiments. The results demonstrate that the proposed approach achieves state-of-the-art performance for bounding boxes at all precision levels and is robust to mild and moderate errors in the loose bounding box annotations. The codes are available at https://github.com/wangjuan313/wsis-beyond-tightBB.

Fuzzy Attention U-Net Architecture Based Localization and YOLOv5 Detection for Fetal Cardiac Ultrasound Images

The importance of early identification of congenital heart disease is highlighted by the fact that it accounts for around 28% of all congenital abnormalities this is the primary reason why fetus die. The necessity of having a thorough understanding of normal cardiac architecture has been highlighted by the quick advancements in fetal heart imaging techniques that have occurred in recent years. Without this information, it is challenging, even impossible, to distinguish the many manifestations of congenital heart illness. This research suggests an immediate fetal cardiac identification technique employing US pictures with the You Only Look Once v5 (YOLOv5) framework and localization using Fuzzy Attention U-Net (FAU-Net) framework in order to enhance the interpretation of the anatomy of the fetal heart through Ultrasound (US) for precise and instantaneous diagnoses. Localization is accomplished using a FU-Net architecture, which limits the training set to image-level plane labels. This is a crucial component of the proposed study since, for big datasets, it would take too much time to produce bounding box annotations, which are not always captured. The FAU-Net design has been optimized for best performance. The YOLOv5 framework is built to function in real-time and deliver the best results for object identification. With the aid of appropriate fine-tuning, it may function effectively to automatically identify tiny fetal cardiac objects in a fast phase. Depending on how much of a detection overlapped with ground-truth bounding boxes, it was determined if it was a genuine positive or a false positive. This research primarily aids medical professionals in the diagnosis of fetal cardiac anatomy. The efficiency of the outcomes is evaluated using metrics including accuracy, memory, average precision (AP), mean average precision (mAP), and F-measure.

Object-level benchmark for deep learning-based detection and classification of weed species

Weeds can decrease yields and the quality of crops. Detection, localisation, and classification of weeds in crops are crucial for developing efficient weed control and management systems. Deep learning (DL) based object detection techniques have been applied in various applications. However, such techniques generally need appropriate datasets. Most available weed datasets only offer image-level annotation, i.e., each image is labelled with one weed species. However, in practice, one image can have multiple weed (and crop) species and/or multiple instances of one species. Consequently, the lack of instance-level annotations of the weed datasets puts a constraint on the applicability of powerful DL techniques. In the current research, we construct an instance-level labelled weed dataset. The images are sourced from a publicly available weed dataset, namely the Corn weed dataset. It has 5997 images of Corn plants and four types of weeds. We annotated the dataset using a bounding box around each instance and labelled them with the appropriate species of the crop or weed. Overall, the images contain about three bounding box annotations on average, while some images have over fifty bounding boxes. To establish the benchmark dataset, we evaluated the dataset using several DL models, including YOLOv7, YOLOv8 and Faster-RCNN, to locate and classify weeds in crops. The performance of the models was compared based on inference time and detection accuracy. YOLOv7 and its variant YOLOv7-tiny models both achieved the highest mean average precision (mAP) of 88.50% and 88.29% and took 2.7 and 1.43 ms, respectively, to classify crop and weed species in an image. YOLOv8m, a variant of YOLOv8, detected the plants in 2.2 ms with the mAP of 87.75%. Data augmentation to address the class imbalance in the dataset improves the mAP results to 89.93% for YOLOv7 and 89.39% for YOLOv8. The detection accuracy and inference time performed by YOLOv7 and YOLOv8 models in this research indicate that these techniques can be used to develop an automatic field-level weed detection system.

A method of concrete damage detection and localization based on weakly supervised learning

AbstractAutomatic inspection of concrete surface defects based on visual elements is crucial for the timely detection of security risks in infrastructure. Moreover, accurate determination of the geographical location of the detected defects is critical for subsequent maintenance and reinforcement tasks. This study employed convolutional neural network (CNN) training methods for detection and localization. This approach employs bounding boxes to confine damaged pixels and utilizes projection loss to foster similarity learning between pixels. In addition, geotags are automatically indexed through the vectorization of invariant features in a scene, which maintains high model accuracy and reduces training costs. The proposed method can detect and classify typical concrete defects in complex scenarios and accurately locate them without the use of external sensors. In addition, the proposed model can achieve pixel‐level defect detection and geographic location determination through the cost of bounding box annotation and automatic indexing. The proposed model was evaluated using 10,691 images of four typical concrete defects in various complex environments. The results demonstrate that the proposed method achieves a pixel‐level detection accuracy of 48.75 and a location accuracy of 83.69, showing a better performance than other methods.

Characterizing Time Spent in Video Object Tracking Annotation Tasks: A Study of Task Complexity in Vehicle Tracking

Video object tracking annotation tasks are a form of complex data labeling that is inherently tedious and time-consuming. Prior studies of these tasks focus primarily on quality of the provided data, leaving much to be learned about how the data was generated and the factors that influenced how it was generated. In this paper, we take steps toward this goal by examining how human annotators spend their time in the context of a video object tracking annotation task. We situate our study in the context of a standard vehicle tracking task with bounding box annotation. Within this setting, we study the role of task complexity by controlling two dimensions of task design -- label constraint and label granularity -- in conjunction with worker experience. Using telemetry and survey data collected from 40 full-time data annotators at a large technology corporation, we find that each dimension of task complexity uniquely affects how annotators spend their time not only during the task, but also before it begins. Furthermore, we find significant misalignment in how time-use was observed and how time-use was self-reported. We conclude by discussing the implications of our findings in the context of video object tracking and the need to better understand how productivity can be defined in data annotation.

Tryp: a dataset of microscopy images of unstained thick blood smears for trypanosome detection

Trypanosomiasis, a neglected tropical disease (NTD), challenges communities in sub-Saharan Africa and Latin America. The World Health Organization underscores the need for practical, field-adaptable diagnostics and rapid screening tools to address the negative impact of NTDs. While artificial intelligence has shown promising results in disease screening, the lack of curated datasets impedes progress. In response to this challenge, we developed the Tryp dataset, comprising microscopy images of unstained thick blood smears containing the Trypanosoma brucei brucei parasite. The Tryp dataset provides bounding box annotations for tightly enclosed regions containing the parasite for 3,085 positive images, and 93 images collected from negative blood samples. The Tryp dataset represents the largest of its kind. Furthermore, we provide a benchmark on three leading deep learning-based object detection techniques that demonstrate the feasibility of AI for this task. Overall, the availability of the Tryp dataset is expected to facilitate research advancements in diagnostic screening for this disease, which may lead to improved healthcare outcomes for the communities impacted.