Fig 1.
Overview of the YOLOv5 Architecture.
Fig 2.
The architecture of the proposed LP-GAN.
Table 1.
Architecture of the generator in LP-GAN.
Table 2.
Architecture of the discriminator in LP-GAN.
Fig 3.
Evolution of generator outputs during training on (a) rainy, (b) foggy and (c) nighttime scenes.
Fig 4.
Visual results of rainy-to-clear image translation using LP-GAN, with each column showing (from top to bottom) the rainy input, the generated output, and the clear-weather ground truth.
Fig 5.
Visual results of foggy-to-clear image translation using LP-GAN, with each column showing (from top to bottom) the foggy input, the generated output, and the clear-weather ground truth.
Fig 6.
Visual results of nighttime-to-clear image translation using LP-GAN, with each column showing (from top to bottom) the nighttime input, the generated output, and the clear-weather ground truth.
Fig 7.
Comparative performance of LP-GAN, standard GAN, and AE on (a) rainy, (b) foggy, and (c) nighttime scenes.
Fig 8.
Object detection results in three weather conditions.
Fig 9.
mAP comparison under three different adverse weather conditions.