SlimYOLOv3: Narrower, Faster and Better for Real-Time UAV Applications

Abstract

Drones or general Unmanned Aerial Vehicles (UAVs), endowed with computer vision function by onboard cameras and embedded systems, have become popular in a wide range of applications. However, real-time scene parsing through object detection running on a UAV platform is very challenging, due to limited memory and computing power of embedded devices. To deal with these challenges, in this paper we propose to learn efficient deep object detectors through channel pruning of convolutional layers. To this end, we enforce channel-level sparsity of convolutional layers by imposing L1 regularization on channel scaling factors and prune less informative feature channels to obtain "slim" object detectors. Based on such approach, we present SlimYOLOv3 with fewer trainable parameters and floating point operations (FLOPs) in comparison of original YOLOv3 (Joseph Redmon et al., 2018) as a promising solution for real-time object detection on UAVs. We evaluate SlimYOLOv3 on VisDrone2018-Det benchmark dataset; compelling results are achieved by SlimYOLOv3 in comparison of unpruned counterpart, including ~90.8% decrease of FLOPs, ~92.0% decline of parameter size, running ~2 times faster and comparable detection accuracy as YOLOv3. Experimental results with different pruning ratios consistently verify that proposed SlimYOLOv3 with narrower structure are more efficient, faster and better than YOLOv3, and thus are more suitable for real-time object detection on UAVs. Our codes are made publicly available at https://github.com/PengyiZhang/SlimYOLOv3. Figure 1. Billion floating point operations (BFLOPs) versus accuracy (mAP) on VisDrone2018-Det benchmark dataset. Enabled by channel pruning, our SlimYOLOv3-SPP3 achieves comparable detection accuracy as YOLOv3 but only requires the equivalent floating point operations as YOLOv3-tiny. Such performance is very competitive in drone applications. Details are given in §5.

Collaborative Multi-Robot Search and Rescue: Planning, Coordination, Perception, and Active Vision

“…Object detection tasks require high computing power and memory for real-time applications. Therefore, cloud computing [198] or small-sized object detection methods have been used for UAV applications [199]- [202]. Cloud computing assists the system with high computing power and memory.…”

“…Another option is to rely on specific object detection mod-els [199]- [202], designed for limited computational power and memory. The papers proposed new object detection models, by using old detection models as their base structure and scaling the original network by reducing the number of filters or changing the layers and they achieved comparable detection accuracy besides the speed on real-time applications on drones.…”

Mixed YOLOv3-LITE: A Lightweight Real-Time Object Detection Method

“…We selected Trial 17 (i.e., Mixed YOLOv3-LITE), which yielded the best results using the PASCAL VOC dataset, to train on the VisDrone 2018 dataset. Sixty epochs of training were performed using the training set with input image data of size 832 × 832, tested using the validation dataset, and compared with the data for SlimYOLOv3 [34]. The results are shown in Table 7.…”

“…The data category distribution of the VisDrone2018-Det dataset is highly uneven, which is more challenging. For example, instances of Note: Tiny-YOLOv3 and SlimYOLOv3 series network FPS data were measured in the NVIDIA GTX1080Ti environment used in Reference [34].…”

“…Note: Tiny-YOLOv3 and SlimYOLOv3 series network FPS data were measured in the NVIDIA GTX1080Ti environment used in Reference[34].…”