Drone detection has become an essential task in object detection as drone costs have decreased and drone technology has improved. It is, however, difficult to detect distant drones when there is weak contrast, long range, and low visibility. In this work, we propose several sequence classification architectures to reduce the detected false-positive ratio of drone tracks. Moreover, we propose a new drone vs. bird sequence classification dataset to train and evaluate the proposed architectures. 3D CNN, LSTM, and Transformer based sequence classification architectures have been trained on the proposed dataset to show the effectiveness of the proposed idea. As experiments show, using sequence information, bird classification and overall F1 scores can be increased by up to 73% and 35%, respectively. Among all sequence classification models, R(2+1)D-based fully convolutional model yields the best transfer learning and fine-tuning results.
Range-gated imaging systems are active systems which use a high-power pulsed-light source and control the opening and closing times of the camera shutter in conjunction with the light source. By calculating the arrival time of the reflected light from the object, the camera shutter is opened for a short time period to form an image using the returned light. This allows generating high contrast images of the objects in difficult lighting conditions. On the other hand the object distance needs to be known and operators are expected to select the proper shutter timing to keep the object of interest continuously in the view. In order to automate this procedure, a tracking system needs to provide feedback to adjust camera shutter timing by estimating the distance of the object in addition to its horizontal and vertical position. In this paper, we present an object tracking framework integrated to the range-gated camera setup without resorting to an additional laser or radar based range finder unit even the object distance changes during the tracking. Range estimation is solely based on image processing and the distance of the object is estimated by the proposed algorithm with a number of similarity measurement methods. The performances of these methods are compared for various scenarios using the data acquired by the range-gated system setup.
Automatic detection and tracking of objects get more important with the increasing number of surveillance cameras and
mobile platforms having cameras. Tracking systems are either designed with stationary camera or designed to work in
moving camera. When the camera is stationary, correspondence based tracking with background subtraction has a
number of benefits such as enabling automatic detection of new objects in the scene and better tracking accuracy. On the
other hand, mean shift is a histogram-based tracking method which is suitable for tracking objects under unconstrained
scenarios like moving camera. However, with mean shift, the objects to be tracked cannot be detected automatically,
only existing or manually selected objects can be tracked. In this paper, we propose a dual-mode system which combines
the advantages of correspondence based tracking and mean shift tracking. A reliability measure based on background
update rate is calculated for each frame. Under normal operating conditions, when the background estimation is working
reliably, correspondence based tracking is used. When the reliability of background estimation becomes low, due to
moving camera, the system automatically switches to mean shift tracking until the reliability of background information
increases again. The results show that the system can detect new objects and track them reliably using background
subtraction. Even though the background subtraction based systems detect high number of false objects when the camera
starts moving, the proposed system hands over the tracked objects to mean shift tracker and avoids detection of false
objects and enables uninterrupted tracking.
Timely detection of packages that are left unattended in public spaces is a security concern, and rapid detection is important for prevention of potential threats. Because constant surveillance of such places is challenging and labor intensive, automated abandoned-object-detection systems aiding operators have started to be widely used. In many studies, stationary objects, such as people sitting on a bench, are also detected as suspicious objects due to abandoned items being defined as items newly added to the scene and remained stationary for a predefined time. Therefore, any stationary object results in an alarm causing a high number of false alarms. These false alarms could be prevented by classifying suspicious items as living and nonliving objects. In this study, a system for abandoned object detection that aids operators surveilling indoor environments such as airports, railway or metro stations, is proposed. By analysis of information from a thermal- and visible-band camera, people and the objects left behind can be detected and discriminated as living and nonliving, reducing the false-alarm rate. Experiments demonstrate that using data obtained from a thermal camera in addition to a visible-band camera also increases the true detection rate of abandoned objects.
Mycotoxins are toxic secondary metabolites produced by fungi. They have been demonstrated to cause various health
problems in humans, including immunosuppression and cancer. A class of mycotoxins, aflatoxins, has been studied
extensively because they have caused many deaths particularly in developing countries. Chili pepper is also prone to
aflatoxin contamination during harvesting, production and storage periods. Chemical methods to detect aflatoxins are
quite accurate but expensive and destructive in nature. Hyperspectral and multispectral imaging are becoming
increasingly important for rapid and nondestructive testing for the presence of such contaminants. We propose a compact
machine vision system based on hyperspectral imaging and machine learning for detection of aflatoxin contaminated
chili peppers. We used the difference images of consecutive spectral bands along with individual band energies to
classify chili peppers into aflatoxin contaminated and uncontaminated classes. Both UV and halogen illumination
sources were used in the experiments. The significant bands that provide better discrimination were selected based on
their neural network connection weights. Higher classification rates were achieved with fewer numbers of spectral bands.
This selection scheme was compared with an information-theoretic approach and it demonstrated robust performance
with higher classification accuracy.
Separate tracking of objects such as people and the luggages they carry is important for video surveillance applications
as it would allow making higher level inferences and timely detection of potential threats. However, this is a challenging
problem and in the literature, people and objects they carry are tracked as a single object. In this study, we propose using
thermal imagery in addition to the visible band imagery for tracking in indoor applications (such as airports, metro or
railway stations). We use adaptive background modeling in association with mean-shift tracking for fully automatic
tracking. Trackers are refreshed using the background model to handle occlusion and split and to detect newly emerging
objects as well as objects that leave the scene. Visible and thermal domain tracking information are fused to allow
tracking of people and the objects they carry separately using their heat signatures. By using the trajectories of these
objects, interactions between them could be deduced and potential threats such as abandoning of an object by a person
could be detected in real-time. Better tracking performance is also achieved compared to using a single modality as
thermal reflection and halo effect which adversely affect tracking are eliminated by the complementing visible band data.
The proposed method has been tested on videos containing various scenarios. The experimental results show that the
presented method is effective for separate tracking of objects such as people and their belongings and for detecting the
interactions in the presence of occlusions.
Satellite images captured in different spectral bands might exhibit nonlinear intensity changes at the corresponding spatial locations due to the different reflectance responses for these bands. This affects the image registration performance negatively as the corresponding features might have different properties in different bands. We propose a modification to the widely used scale invariant feature transform (SIFT) method to increase the correct feature matching ratio and to decrease the computation time of this algorithm for the multispectral satellite images. We also apply scale restriction to SIFT and speeded up robust features (SURF) algorithms to increase the correct match ratio. We present test results for variations of SIFT and SURF algorithms. The results show the effectiveness of the proposed improvements compared to the other SIFT- and SURF-based methods.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.