Search

Vehicle detection in aerial images is an interesting and challenging task. Traditional methods are based on sliding‐window search and handcrafted features, which limits the representation power and has heavy computational costs. Recent research have shown that deep‐learning algorithms are widely used in the field of object detection. However, the deep‐learning algorithms still face many difficulties and challenges in the object detection under the aerial scene. Meanwhile, the high computational cost of detection algorithms lead to low‐detection efficiency. In this study, we build a fast and accurate lightweight detection framework for vehicle detection in aerial scenes. The proposed detection method improves the expressive ability of detection network and significantly reduces the amount of calculations in the model. Meanwhile, setting suitable anchor boxes according to the size of the object vehicles have been introduced in our model, which also effectively improves the performance of the detection. In addition, we have published a new aerial vehicle image dataset and verified the effectiveness of our method. In the Munich dataset and our dataset, our method achieves 85.8% and 91.2% of the mean average precision (mAP), and its detection time is 1.78 and 0.048 s on Nvidia Titan XP. Our results show that the proposed framework achieves significant improvement over several alternatives and state‐of‐the‐art schemes with higher accuracy and less detection time. [ABSTRACT FROM AUTHOR]

Purpose: To correlate retinal ganglion cell (RGC) loss and optic nerve (ON) damage with the duration of acute glaucoma attacks in a rat experimental model and to determine whether the c-Jun N-terminal kinase (JNK) inhibitor SP600125 protects against such attacks., Methods: To model an acute glaucoma attack, rat intraocular pressure (IOP) was elevated by a controllable compression method using pulleys and specific weights. Intraocular pressure was measured with a TonoLab® rebound tonometer. Time-dependent ocular hypertension-induced damage was evaluated by ON morphology, retina morphology (both retina layer thickness in cross-sections and RGC counts in Dextran tetramethylrhodamine crystals [DTMR] labeled flatmounts), and scotopic flash electroretinography (ERG). A c-Jun N-terminal kinase (JNK) inhibitor, SP600125 (0, 1.5, 5, or 15 mg/kg), was administered by intraperitoneal injection immediately before and after induction of ocular hypertension, then once daily for seven days. Retinal cross-sections were measured to determine the thickness of various retinal layers and the cell density in the ganglion cell layer (GCL). Retinal flatmounts immunolabeled with anti-rat Brn-3a primary antibody were used to quantify RGC numbers., Results: Elevated rat IOP induced by corneal limbus compression correlated with the different weights. Elevation to 45 mmHg for up to 7 h did not significantly affect the thicknesses of the outer nuclear layer, outer plexiform layer, or inner nuclear layer. Amplitudes of A- and B-waves were not affected. However, elevation to 45 mmHg for up to 7 h decreased the inner retinal thickness and caused ON damage. Most importantly, IOP elevation induced a time-dependent RGC loss. Cell density in the GCL decreased to 70%, 62%, and 49% of that of the control after 5 h, 6 h, and 7 h, respectively, of pressure increases. In retinal flatmount studies, labeled RGCs were reduced 56±4% (mean±SEM) versus the control (p<0.001) after 7 h of ocular hypertension. SP600125 dose-dependently protected against ocular hypertension-induced RGC loss. The difference in RGC density between the vehicle and SP600125-treated (15 mg/kg) groups was statistically significant (p<0.001)., Conclusions: The correlation of inner retinal morphological changes with the duration of the application of 45 mmHg IOP was demonstrated. Treatment with SP600125 significantly protected RGC survival against this insult. Inhibitors of JNK may be an interesting pharmacological class for treating glaucoma.