Search

Adopting effective techniques to automatically detect and identify small drones is a very compelling need for a number of different stakeholders in both the public and private sectors. This work presents three different original approaches that competed in a grand challenge on the “Drone vs. Bird” detection problem. The goal is to detect one or more drones appearing at some time point in video sequences where birds and other distractor objects may be also present, together with motion in background or foreground. Algorithms should raise an alarm and provide a position estimate only when a drone is present, while not issuing alarms on birds, nor being confused by the rest of the scene. In particular, three original approaches based on different deep learning strategies are proposed and compared on a real-world dataset provided by a consortium of universities and research centers, under the 2020 edition of the Drone vs. Bird Detection Challenge. Results show that there is a range in difficulty among different test sequences, depending on the size and the shape visibility of the drone in the sequence, while sequences recorded by a moving camera and very distant drones are the most challenging ones. The performance comparison reveals that the different approaches perform somewhat complementary, in terms of correct detection rate, false alarm rate, and average precision.

Much of the focus in the object detection literature has been on the problem of identifying the bounding box of a particular class of object in an image. Yet, in contexts such as robotics and augmented reality, it is often necessary to find a specific object instance---a unique toy or a custom industrial part for example---rather than a generic object class. Here, applications can require a rapid shift from one object instance to another, thus requiring fast turnaround which affords little-to-no training time. What is more, gathering a dataset and training a model for every new object instance to be detected can be an expensive and time-consuming process. In this context, we propose a generic 2D object instance detection approach that uses example viewpoints of the target object at test time to retrieve its 2D location in RGB images, without requiring any additional training (i.e. fine-tuning) step. To this end, we present an end-to-end architecture that extracts global and local information of the object from its viewpoints. The global information is used to tune early filters in the backbone while local viewpoints are correlated with the input image. Our method offers an improvement of almost 30 mAP over the previous template matching methods on the challenging Occluded Linemod dataset (overall mAP of 50.7). Our experiments also show that our single generic model (not trained on any of the test objects) yields detection results that are on par with approaches that are trained specifically on the target objects.

Accurate pose estimation is often a requirement for robust robotic grasping and manipulation of objects placed in cluttered, tight environments, such as a shelf with multiple objects. When deep learning approaches are employed to perform this task, they typically require a large amount of training data. However, obtaining precise 6 degrees of freedom for ground-truth can be prohibitively expensive. This work therefore proposes an architecture and a training process to solve this issue. More precisely, we present a weak object detector that enables localizing objects and estimating their 6D poses in cluttered and occluded scenes. To minimize the human labor required for annotations, the proposed detector is trained with a combination of synthetic and a few weakly annotated real images (as little as 10 images per object), for which a human provides only a list of objects present in each image (no time-consuming annotations, such as bounding boxes, segmentation masks and object poses). To close the gap between real and synthetic images, we use multiple domain classifiers trained adversarially. During the inference phase, the resulting class-specific heatmaps of the weak detector are used to guide the search of 6D poses of objects. Our proposed approach is evaluated on several publicly available datasets for pose estimation. We also evaluated our model on classification and localization in unsupervised and semi-supervised settings. The results clearly indicate that this approach could provide an efficient way toward fully automating the training process of computer vision models used in robotics.

Pick-and-place is an important task in robotic manipulation. In industry, template-matching approaches are often used to provide the level of precision required to locate an object to be picked. However, if a robotic workstation is to handle numerous objects, brute-force template-matching becomes expensive, and is subject to notoriously hard-totune thresholds. In this paper, we explore the use of Deep Learning methods to speed up traditional methods such as template matching. In particular, we employed a Single Shot Detection (SSD) and a Residual Network (ResNet) for object detection and classification. Classification scores allows the re-ranking of objects so that template matching is performed in order of likelihood. Tests on a dataset containing 10 industrial objects demonstrated the validity of our approach, by getting an average ranking of 1.37 for the object of interest. Moreover, we tested our approach on the standard Pose dataset which contains 15 objects and got an average ranking of 1.99. Because SSD and ResNet operates essentially in constant time in a Graphics Processor Unit, our approach is able to reach near-constant time execution. We also compared the F1 scores of LINE-2D, a state-of-the-art template matching method, using different strategies (including our own) and the results show that our method is competitive to a brute-force template matching approach. Coupled with near-constant time execution, it therefore opens up the possibility for performing template matching for databases containing hundreds of objects.

This paper proposes a complete system for robotic sensor placement in initially unknown arbitrary three-dimensional environments. The system uses a novel approach for computing the quality of acquisition of a mobile sensor group in such environments. The quality of acquisition is based on a geometric model of a camera which allows accurate sensor models and simple occlusion computation. The proposed system combines this new metric with a global derivative-free optimization algorithm to find simultaneously the number of sensors and their configuration to sense accordingly the environment. The presented framework compares favourably with current techniques working in two-dimensional environments. Furthermore, simulation and experimental results demonstrate the ability of the system to cope with full three-dimensional environments, a domain still unexplored by previous methods.

BACKGROUND Although low dietary intakes of antioxidant vitamins and minerals have been associated with higher risks of cancer, results of trials testing antioxidant supplementation for cancer chemoprevention have been equivocal. We assessed whether supplementation with antioxidant vitamins could reduce the incidence of second primary cancers among patients with head and neck cancer. METHODS We conducted a multicenter, double-blind, placebo-controlled, randomized chemoprevention trial among 540 patients with stage I or II head and neck cancer treated by radiation therapy between October 1, 1994, and June 6, 2000. Supplementation with alpha-tocopherol (400 IU/day) and beta-carotene (30 mg/day) or placebo began on the first day of radiation therapy and continued for 3 years after the end of radiation therapy. In the course of the trial, beta-carotene supplementation was discontinued after 156 patients had enrolled because of ethical concerns. The remaining patients received alpha-tocopherol or placebo only. Survival was evaluated by Kaplan-Meier analysis. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). All statistical tests were two-sided. RESULTS After a median follow-up of 52 months, second primary cancers and recurrences of the first tumor were diagnosed in 113 and 119 participants, respectively. The effect of supplementation on the incidence of second primary cancers varied over time. Compared with patients receiving placebo, patients receiving alpha-tocopherol supplements had a higher rate of second primary cancers during the supplementation period (HR = 2.88, 95% CI = 1.56 to 5.31) but a lower rate after supplementation was discontinued (HR = 0.41, 95% CI = 0.16 to 1.03). Similarly, the rate of having a recurrence or second primary cancer was higher during (HR = 1.86, 95% CI = 1.27 to 2.72) but lower after (HR = 0.71, 95% CI = 0.33 to 1.53) supplementation with alpha-tocopherol. The proportion of participants free of second primary cancer overall after 8 years of follow-up was similar in both arms. CONCLUSIONS alpha-Tocopherol supplementation produced unexpected adverse effects on the occurrence of second primary cancers and on cancer-free survival.

Anion exchange chromatography with evaporative light scattering detection was investigated for the direct and simultaneous analysis of methyl and ethylphosphonic acids and inorganic anions in aqueous media. The chromatographic system made up of a Dionex IonPac AS4A-SC column and an aqueous mobile phase containing NH4HCO3 25 mM (pH 7.8) allows the resolution of the two phosphonic acids in less than 10 minutes without coelution of chloride, phosphate, and sulfate. This methodology has been successfully applied to the direct analysis of aqueous matrix containing low levels of phosphonic acids and high levels of inorganic anions (chloride 1700 mg/L, phosphate 1000 mg/L, or sulfate 1200 mg/L) without any baseline perturbation and matrix pretreatment. Quantitative analysis of phosphonic acids is achieved with a good correlation coefficient, and a 200 ng (200 μL of 1 mg/L) detection limit of MPA or EPA is obtainable.

Purpose To determine whether efaproxiral, an allosteric modifier of hemoglobin, improves survival in patients with brain metastases when used as an adjunct to whole-brain radiation therapy (WBRT). Patients and Methods Patients with brain metastases from solid tumors and a Karnofsky performance score of ≥ 70 were randomly assigned to receive WBRT with supplemental oxygen and either efaproxiral at 75 or 100 mg/kg (efaproxiral arm) or no efaproxiral (control arm). The primary end point was survival. Results The study consisted of 515 eligible patients (efaproxiral arm, n = 265; control arm, n = 250). The median survival time (MST) was 5.4 months for the efaproxiral arm versus 4.4 months for the control arm (hazard ratio [HR] = 0.87; P = .16). For the subgroup of patients with non–small-cell lung cancer (NSCLC) or breast cancer, the MST was 6.0 and 4.4 months, respectively (HR = 0.82; P = .07). Cox multiple regression analysis demonstrated a significant reduction in the risk of death for the efaproxiral arm in both primary populations. Further analysis indicated that the benefit may be restricted to the subgroup of patients with breast cancer. Response rates (radiographic complete response plus partial response) improved by 7% (P = .10) and 13% (P = .01) for all patients and for NSCLC and breast cancer patients in the efaproxiral arm, respectively. The most common severe adverse event in patients treated with efaproxiral was hypoxemia, which was reversible and effectively managed with supplemental oxygen in most patients. Conclusion The addition of efaproxiral, a noncytotoxic radiation sensitizer, to WBRT may improve response rates and survival in patients with brain metastases, particularly metastases from breast cancer. A confirmatory trial for breast cancer patients has been initiated.

Purpose Many cancer patients take antioxidant vitamin supplements with the hope of improving the outcome of conventional therapies and of reducing the adverse effects of these treatments. A randomized trial was conducted to determine whether supplementation with antioxidant vitamins could reduce the occurrence and severity of acute adverse effects of radiation therapy and improve quality of life without compromising treatment efficacy. Patients and Methods We conducted a randomized, double-blind, placebo-controlled trial among 540 head and neck cancer patients treated with radiation therapy. Patients were randomly assigned into two arms. The supplementation with α-tocopherol (400 IU/d) and β-carotene (30 mg/d) or placebos was administered during radiation therapy and for 3 years thereafter. During the course of the trial, supplementation with β-carotene was discontinued because of ethical concerns. Results Patients randomly assigned in the supplement arm tended to have less severe acute adverse effects during radiation therapy (odds ratio [OR], 0.72; 95% CI, 0.52 to 1.02). The reduction was statistically significant when the supplementation combined α-tocopherol and β-carotene for adverse effects to the larynx (OR, 0.38; 95% CI, 0.21 to 0.71) and overall at any site (OR, 0.38; 95% CI, 0.20 to 0.74). Quality of life was not improved by the supplementation. The rate of local recurrence of the head and neck tumor tended to be higher in the supplement arm of the trial (hazard ratio, 1.37; 95% CI, 0.93 to 2.02). Conclusion Supplementation with high doses of α-tocopherol and β-carotene during radiation therapy could reduce the severity of treatment adverse effects. However, this trial suggests that use of high doses of antioxidants as adjuvant therapy might compromise radiation treatment efficacy.

1534 Background: Brain metastases represent a common cause of morbidity and mortality among cancer pts. The effectiveness of WBRT as primary therapy for brain metastases is limited by tissue hypoxi...