Unmanned Aerial Vehicle Flies Research Articles

ERROR ANALYSIS OF VISUAL ODOMETRY FOR A SMALL SIZE UNMANNED AERIAL VEHICLE

Abstract. Usage of small size Unmanned Aerial Vehicle (UAV) with autonomous flight capability became widespread at both civil and military applications because of its low costs and ease of use. However, inertial or GPS based positioning and orientation sensors can be disabled at battlefields and the UAV losses its path in a short time. Visual odometry becomes a solution for the aforementioned hassle. Real-time detection of ground control points, matching the same control points at the stereo pair images and correcting the position and orientation data with least square adjustment are the challenges of visual odometry. In this study, error analysis of position and orientation of a UAV flying through mountainous region which cannot obtain any augmenting data for position and orientation is conducted by error propagation law. The flight of UAV is simulated with exact initial position and orientation data. Optic images of ground obtained by nadir directed camera are georeferenced using the digital elevation model of the region, interior and exterior camera parameters. The digital elevation model is used to correct the relief displacement. The stereo-pair image is obtained when the UAV flies 200 m and image coordinates of the common control points are detected by colinearity equations. Position and orientation of UAV is corrected by least square adjustment and the process is repeated for each obtained image acquired at 200 m interval. The analysis revealed that the precisions of horizontal and vertical positioning becomes 75 and 200 meter, respectively. The precision of roll angle becomes worse than 0.5 degree, while the precisions of pitch and yaw angle are estimate to be better that 0.1 degree.

Traffic offloading for postdisaster rescue in UAV‐assisted networks: A coalition formation game approach

SummaryIn postdisaster rescue scenario, unmanned aerial vehicles (UAVs) are effective tools to help ground users in disaster areas to transmit rescue‐critical data to the relief center in time due to their flexible mobilities and fast deployment. However, how UAVs choose disaster areas to take part in traffic offloading with limited bandwidth resources is challenging. This article investigates UAVs' access selection and disaster areas' bandwidth allocation scheme by jointly optimizing network throughput, bandwidth and UAVs' energy cost. The UAVs form coalitions to participate in traffic offloading cooperatively, where each UAV selects a disaster area independently. Specifically, the bandwidth resources are dynamically allocated to all disaster areas when a new coalition partition forms. The UAVs adjust their access choices for disaster areas when the bandwidth allocation changes. When no UAVs migrate to other disaster areas, the network achieves a stable state. The UAVs fly to the final selected disaster area and provide traffic offloading services. To resolve traffic offloading competitions yet enable cooperations among UAVs, we address the UAVs' access selection issue by coalition formation game. A gradient projection method is then proposed to allocate bandwidth resources which maximizes the benefit of the network. We demonstrate that the UAVs' access selection and disaster areas' bandwidth allocation algorithms are convergent. The simulation results demonstrate that our proposed cooperation order is better than the pareto and selfish orders, thereby increasing the benefit of the network.

Physical layer security‐oriented energy‐efficient resource allocation and trajectory design for UAV jammer over 5G millimeter wave cognitive relay system

AbstractThis work focuses on the optimal trajectory of a rotary‐wing unmanned aerial vehicle (UAV) jammer and the system power allocation with the objective of maximizing the total secure energy efficiency (EE) of a slotted relay‐assisted millimeter wave (mm‐Wave) cognitive radio system. The UAV flies from starting location to final one in a round of communication mission that is overheard by a malicious eavesdropper. The network terminals are equipped with multiple‐input multiple‐output (MIMO) antenna array. This work first develops a practical approximation propagation model by addressing three key challenges: path loss, small‐scale Nakagami‐m fading channel gains, and transceiver antenna beamforming gains. By using the approximate model, the upper and lower bounds of the ergodic rate of each communication link are derived. Second, with the objective of maximizing secure EE, the optimal UAV trajectory and system resource allocation framework is established by considering the constraints of interference at primary receiver (PR), information‐causality relationship, rotary‐wing UAV propulsion energy and so forth. The optimal framework is solved with a tractable CVX tool by using the derived upper and lower bounds and successive convex approximation. Finally, an iterative optimizing algorithm is established by using Dinkelbach's method. In the simulation, the work presents the comparison of UAV trajectories, secure rates, and secure EEs of three schemes, that is, the proposed EE scheme, the non‐EE scheme where only the optimal secure rate is considered, and the EE scheme with the consideration of all communication‐related energy consumption. The simulations show that the EE schemes achieve smoother UAV trajectory than non‐EE, but the non‐EE scheme has higher secure rate. The impact of communication energy consumption is negligible. In addition, the effect of MIMO antenna array and mm‐Wave on system performance is exploited.

Energy optimization for full-duplex wireless-powered IoT networks using rotary-wing UAV with multiple antennas

In this paper, we propose a novel design for the rotary-wing unmanned aerial vehicle (UAV)-enabled full-duplex (FD) wireless-powered Internet of Things (IoT) networks. In this network, the UAV is equipped with an antenna array, and the K IoT sensors, which are distributed randomly, use single-antenna to communicate. By sending the energy, the UAV as a hybrid access point, charges the sensors and collects information from them. Then, to manage the time and optimize the energy, the sensors are divided into N groups, so that the UAV equipped with multi-input multi-output (MIMO) technology can serve the sensors in a group, during the total time T. We provide a simple implementation of the wireless power transfer protocol in the sensors by using the time division multiple access (TDMA) scheme to receive information from the users. In other words, the sensors of each group receive energy from the UAV, when it hovers over the sensors of the previous group, and also when the UAV flies over the previous group to the current group. The sensors of each group send their information to the UAV, when the UAV is hovering over their group. Under these assumptions, we formulate two optimization problems: a sum throughput maximization problem, and a total time minimization problem. Numerical results show that our proposed optimal network provides better performance than the existing networks. In fact, our novel design can serve more sensors at the cost of using more antennas compared to that of the conventional networks.

Energy Minimization for UAV-Enabled Wireless Power Transfer and Relay Networks

In this paper, we consider an unmanned aerial vehicle (UAV)-enabled wireless power transfer (WPT) and relay communication network consisting of a base station (BS), a UAV, and multiple ground users. The UAV acts as both a wireless power transmission source and an uplink communication relay. Specifically, an entire transmission period of the considered system is divided into two stages. In the first stage, the UAV transfers the power to the ground users along a well-optimized flight trajectory and meanwhile the users transmit data to the UAV using the harvested energy. Subsequently, in the second stage, the UAV flies to the vicinity of the BS and forwards the data to the BS. For the purpose of minimizing the energy consumed by the UAV, we jointly optimize the time durations of the two stages, the UAV’s transmit powers for WPT and data forwarding, as well as its flight trajectory, subject to the constraints of the quality of service (QoS), the information forwarding, the energy causality, and the mobility of the UAV. The involved optimization problem is non-convex and highly intractable. To this end, we propose an efficient alternating algorithm to iteratively solve two subproblems with respect to the time durations of the two stages and the UAV’s transmit powers and trajectory, respectively. The first subproblem has a closed-form optimal solution and the second subproblem is handled by addressing a surrogate convex problem based on the technique of successive convex approximation. Finally, simulation results confirm the superiority of our proposed algorithm.

In-Vehicle Speech Recognition for Voice-Driven UAV Control in a Collaborative Environment of MAV and UAV

Most conventional speech recognition systems have mainly concentrated on voice-driven control of personal user devices such as smartphones. Therefore, a speech recognition system used in a special environment needs to be developed in consideration of the environment. In this study, a speech recognition framework for voice-driven control of unmanned aerial vehicles (UAVs) is proposed in a collaborative environment between manned aerial vehicles (MAVs) and UAVs, where multiple MAVs and UAVs fly together, and pilots on board MAVs control multiple UAVs with their voices. Standard speech recognition systems consist of several modules, including front-end, recognition, and post-processing. Among them, this study focuses on recognition and post-processing modules in terms of in-vehicle speech recognition. In order to stably control UAVs via voice, it is necessary to handle the environmental conditions of the UAVs carefully. First, we define control commands that the MAV pilot delivers to UAVs and construct training data. Next, for the recognition module, we investigate an acoustic model suitable for the characteristics of the UAV control commands and the UAV system with hardware resource constraints. Finally, two approaches are proposed for post-processing: grammar network-based syntax analysis and transaction-based semantic analysis. For evaluation, we developed a speech recognition system in a collaborative simulation environment between a MAV and an UAV and successfully verified the validity of each module. As a result of recognition experiments of connected words consisting of two to five words, the recognition rates of hidden Markov model (HMM) and deep neural network (DNN)-based acoustic models were 98.2% and 98.4%, respectively. However, in terms of computational amount, the HMM model was about 100 times more efficient than DNN. In addition, the relative improvement in error rate with the proposed post-processing was about 65%.

Path optimizing of UAV inspection for 500kV AC transmission line cup-tower based on ant colony algorithm

To ensure unmanned aerial vehicles (UAVs) complete the inspection task of a high-voltage transmission line safely and effectively, it is necessary to consider the interference of a complicated electromagnetic field near a high-voltage transmission line and determine a reasonable inspection distance and path of the UAV. Taking the 500kV EHV AC power line as an instance, firstly, the three-dimensional model of the 500kV ZV1 cup-tower is built, and the finite element method (FEM) is used for computing the electric field and magnetic field of the UAV exterior. According to inspection specifications, the electromagnetic field protection threshold of the UAV is determined, and then the safety area between UAV and the split conductor is determined to be equal to 1.5 m. Combined with the UAV shooting field of view and safety distance, the spatial positions of UAV hovering shooting points are selected to ensure that UAV flies in a safe area. Finally, the path of UAV traversing hovering shooting points is optimized by the ant colony algorithm (ACO), and the optimal path distance is 109.4 m.

Compact and ordered swarms of unmanned aerial vehicles in cluttered environments.

The globally coordinated motion produced by the classical swarm model is typically generated by simple local interactions at the individual level. Despite the success of these models in interpretation, they cannot guarantee compact and ordered collective motion when applied to the cooperation of unmanned aerial vehicle (UAV) swarms in cluttered environments. Inspired by the behavioral characteristics of biological swarms, a distributed self-organized Reynolds (SOR) swarm model of UAVs is proposed. In this model, a social term is designed to keep the swarm in a collision-free, compact, and ordered collective motion, an obstacle avoidance term is introduced to make the UAV avoid obstacles with a smooth trajectory, and a migration term is added to make the UAV fly in a desired direction. All the behavioral rules for agent interactions are designed with as simple a potential function as possible. And the genetic algorithm is used to optimize the parameters of the model. To evaluate the collective performance, we introduce different metrics such as (a) order, (b) safety, (c) inter-agent distance error, (d) speed range. Through the comparative simulation with the current advanced bio-inspired compact and Vasarhelyi swarm models, the proposed approach can guide the UAV swarm to pass through the dense obstacle environment in a safe and ordered manner as a compact group, and has adaptability to different obstacle densities.

Unmanned-aerial-vehicle based optical camera communication system using light-diffusing fiber and rolling-shutter image-sensor.

We put forward and demonstrate a light-diffusing fiber equipped unmanned-aerial-vehicle (UAV) to provide a large field-of-view (FOV) optical camera communication (OCC) system. The light-diffusing fiber can act as a bendable, lightweight, extended and large FOV light source for the UAV-assisted optical wireless communication (OWC). During UAV flying, the light-diffusing fiber light source could be tilted or bended; hence, offering large FOV as well as supporting large receiver (Rx) tilting angle are particularly important for the UAV-assisted OWC systems. To improve the transmission capacity of the OCC system, one method based on the camera shutter mechanism, which is known as rolling-shuttering is utilized. The rolling-shuttering method makes use of the feature of complementary-metal-oxide-semiconductor (CMOS) image sensor to extract signal pixel-row by pixel-row. The data rate can be significantly increased since the capture start time for each pixel-row is different. As the light-diffusing fiber is thin and occupies only a few pixels in the CMOS image frame, Long-short-term-memory neural-network (LSTM-NN) is used to enhance the rolling-shutter decoding. Experimental results show that the light-diffusing fiber can satisfactorily act as an "omnidirectional optical antenna" providing wide FOVs and 3.6 kbit/s can be achieved, accomplishing the pre-forward error correction bit-error-rate (pre-FEC BER = 3.8 × 10-3).

Securing UAV Flying Base Station for Mobile Networking: A Review

A flying base station based on an unmanned aerial vehicle (UAV) uses its mobility to extend its connectivity coverage and improve its communication channel quality to achieve a greater communication rate and latency performances. While UAV flying base stations have been used in emergency events in 5G networking (sporadic and temporary), their use will significantly increase in 6G networking, as 6G expects reliable connectivity even in rural regions and requires high-performance communication channels and line-of-sight channels for millimeter wave (mmWave) communications. Securing the integrity and availability of the base station operations is critical because of the users’ increasing reliance on the connectivity provided by the base stations, e.g., the mobile user loses connectivity if the base station operation gets disrupted. This paper identifies the security issues and research gaps of flying base stations, focusing on their unique properties, while building on the existing research in wireless communications for stationary ground base stations and embedded control for UAV drones. More specifically, the flying base station’s user-dependent positioning, its battery-constrained power, and the dynamic and distributed operations cause vulnerabilities that are distinct from those in 5G and previous-generation mobile networking with stationary ground base stations. This paper reviews the relevant security research from the perspectives of communications (mobile computing, 5G networking, and distributed computing) and embedded/control systems (UAV vehicular positioning and battery control) and then identifies the security gaps and new issues emerging for flying base stations. Through this review paper, we inform readers of flying base station research, development, and standardization for future mobile and 6G networking.

UAV High-Voltage Power Transmission Line Autonomous Correction Inspection System Based on Object Detection

With the development of technology, unmanned aerial vehicles (UAVs) are playing an increasingly important role in the inspection of high-voltage power transmission line. The traditional inspection method relies on the operator to manually control the drone for inspection. Although many companies are using real-time dynamic carrier phase differencing technology to achieve high-precision positioning of UAVs, when UAVs fly autonomously at high altitudes to photograph specific objects, the objects tend to deviate from the center of the picture. To address this error, in this paper, an autonomous UAV inspection system based on object detection is designed. (1) To detect inspection objects, the corresponding dataset is established on the basis of the UAV autonomous inspection task. (2) To obtain the position information of the target object, a lightweight object detector based on the YOLOX network model is designed. First, the backbone is replaced with MobileNetv3. Next, in the neck structure, the Ghost module is introduced and depthwise convolution is applied instead of normal convolution. Then, to embed the location information into the channel attention, coordinate attention is introduced after the output feature layer of the backbone, enabling the lightweight network to operate on a larger area of focus. Finally, to improve the accuracy of the bounding box regression, the α-DIOU loss function is introduced. (3) To obtain the best image acquisition position, the results of object detection combined with the real-time status of the UAV are used. (4) To enable the UAV to complete the final corrections, position control and altitude control are used. Compared to the original YOLOX tiny, the new model improves the mAP_0.5:0.95 metric by about 2 percentage points, with a significant reduction in the number of parameters and computation, while running at 56 FPS on Nvidia NX. This system can effectively solve the problem of the target deviating from the center of the picture when the UAV takes pictures during a high-altitude autonomous inspection, verified by many actual flight experiments.

Dynamic Computation Offloading Based on Q-Learning for UAV-Based Mobile Edge Computing

Emerging mobile edge computing (MEC) can be used in battery-constrained Internet of things (IoT). The execution latency of IoT applications can be improved by offloading computation-intensive tasks to an MEC server. Recently, the popularity of unmanned aerial vehicles (UAVs) has increased rapidly, and UAV-based MEC systems are receiving considerable attention. In this paper, we propose a dynamic computation offloading paradigm for UAV-based MEC systems, in which a UAV flies over an urban environment and provides edge services to IoT devices on the ground. Since most IoT devices are energy-constrained, we formulate our problem as a Markov decision process considering the energy level of the battery of each IoT device. We also use model-free Q-learning for time-critical tasks to maximize the system utility. According to our performance study, the proposed scheme can achieve desirable convergence properties and make intelligent offloading decisions.

Joint Optimization of Service Fairness and Energy Consumption for 3D Trajectory Planning in Multiple Solar-Powered UAV Systems

In this paper, we study the three-dimensional (3D) trajectory optimization problems of unmanned aerial vehicles (UAV) with a solar energy supply, aiming to provide communication coverage for mobile users on the ground. In general, the higher UAVs fly, the more solar energy they collect, but the smaller the range of coverage they could achieve, and vice versa. How to plan optimal trajectories for UAVs so that more users can be encompassed, while allowing UAVs to collect enough solar energy, is a challenging issue. Moreover, we also consider how geographically fair coverage for each ground user can be achieved. To solve these problems, we designed a multiple solar-powered UAV (SP-UAV) energy consumption model and a fairness model, while designed an observation space, state space, action space, and reward function. Then, we proposed a multiple SP-UAV 3D trajectory optimization algorithm based on deep reinforcement learning (DRL). Our algorithm is able to balance the energy consumption of UAVs to extend the system’s lifetime, while avoiding both collisions and flying out of communication range. Finally, we trained our model through simulation experiments and conducted comparative experiments and analysis based on real network topology data. The results show that our algorithm is superior to the existing typical algorithms in coverage, fairness, and lifetime.

Joint Trajectory and Power Optimization for Jamming-Aided NOMA-UAV Secure Networks

The combination of nonorthogonal multiple access (NOMA) technique and unmanned aerial vehicle (UAV) provides an effective solution for achieving massive connections and improving spectrum efficiency. However, the related security risk becomes serious due to the line-of-sight (LoS) channels involved and high transmit power for weaker users in NOMA-UAV networks. In this article, a UAV-assisted NOMA transmission scheme is proposed to achieve secure downlink transmission via artificial jamming, where a UAV flies straightly to serve multiple ground users in the presence of a passive eavesdropper. During the flight, only the closest NOMA users are chosen to connect with the UAV in each time slot to achieve high LoS probability. To balance the security and transmission performance, the tradeoff between the jamming power and the sum rate is investigated by jointly optimizing the power allocation, the user scheduling and the UAV trajectory. The formulated problem is mixed-integer and nonconvex due to the coupled variables. To address this, we first decompose the problem into two subproblems of power allocation and trajectory optimization. Then, they are transformed into convex ones via the first-order Taylor expansion. After that, an iterative algorithm is proposed to solve the convex problem. Finally, numerical results show that the security of the network is well enhanced and verify the effectiveness of the proposed scheme.

Safety Control for Quadrotor UAV Against Ground Effect and Blade Damage

This article proposes a safety control scheme for quadrotor unmanned aerial vehicle (UAV) to counteract ground effect and blade damage. The impacts of ground effect and blade damage are quantitatively analyzed in the established model. The proposed safety control scheme consists of position and attitude loops. Specifically, a proportional and derivative controller is designed as a baseline control with acceleration feedforward in the position loop. A disturbance observer (DO) using the measured Euler angles is exploited to estimate the lumped disturbance forces. The developed DO can effectively mitigate the estimation deviation caused by the angle tracking error. Subsequently, differential flatness theory is used to obtain the Euler angle references and the derivatives. In the attitude loop, sliding mode control and fixed-time sliding mode observer (SMO) are integrated to deal with ground effect and blade damage promptly. To demonstrate the effectiveness of the proposed safety control scheme, flight experiments are carried out, where the quadrotor UAV flies close to the rugged ground in the presence of blade damage.

Joint Deployment Optimization and Flight Trajectory Planning for UAV Assisted IoT Data Collection: A Bilevel Optimization Approach

This work investigates an unmanned aerial vehicle (UAV) assisted IoT system, where a UAV flies to each foothold to collect data from IoT devices, and then return to its start point. For such a system, we aim to minimize the energy consumption by jointly optimizing the deployment and flight trajectory of UAV. It is a mixed-integer non-convex and NP-hard problem. In order to address it, a bilevel optimization approach is proposed, where an upper-level method aims to optimize the deployment of UAV and a lower-level one aims to plan UAV flight trajectory. Specifically, the former optimizes the number and locations of footholds of UAV. This work proposes an improved dandelion algorithm with a novel encoding strategy, in which each dandelion represents a foothold of UAV and the entire dandelion population is seen as an entire deployment. Then, two mutation strategies are designed to adjust the number and locations of footholds. Based on the footholds of the UAV provided by the former, the latter transforms flight trajectory planning into a traveling salesman problem (TSP). This work proposes an iterated greedy algorithm to solve it efficiently. The effectiveness of the proposed bilevel optimization approach is verified on ten instances, and the experimental results show that it significantly outperforms other benchmark approaches.

Small-Object Detection for UAV-Based Images Using a Distance Metric Method

Object detection is important in unmanned aerial vehicle (UAV) reconnaissance missions. However, since a UAV flies at a high altitude to gain a large reconnaissance view, the captured objects often have small pixel sizes and their categories have high uncertainty. Given the limited computing capability on UAVs, large detectors based on convolutional neural networks (CNNs) have difficulty obtaining real-time detection performance. To address these problems, we designed a small-object detector for UAV-based images in this paper. We modified the backbone of YOLOv4 according to the characteristics of small-object detection. We improved the performance of small-object positioning by modifying the positioning loss function. Using the distance metric method, the proposed detector can classify trained and untrained objects through object features. Furthermore, we designed two data augmentation strategies to enhance the diversity of the training set. We evaluated our method on a collected small-object dataset; the proposed method obtained 61.00% mAP50 on trained objects and 41.00% mAP50 on untrained objects with 77 frames per second (FPS). Flight experiments confirmed the utility of our approach on small UAVs, with satisfying detection performance and real-time inference speed.

On performance of full-duplex UAV system with multiple NOMA users and millimeter-wave communications

This article considers millimeter-wave (mmWave) communications in a full-duplex (FD) unmanned aerial vehicle (UAV) system with multiple non-orthogonal multiple access (NOMA) users. In particular, a UAV operates in an FD transmission mode and acts as a decode-and-forward (DF) relay to aid communications between a ground base station and multiple ground users. Unlike previous works on the UAV-aided NOMA systems, where the specific carrier frequencies are often ignored, this work investigates the effect of a specific carrier frequency in the mmWave band and the channel model proposed for the fifth and beyond generations (5G and B5G) of mobile communications. The closed-form expressions of outage probability (OP) and achievable data rate (ADR) of the FD-UAV-NOMA system with mmWave communications (called as the MW-FD-UAV-NOMA system) are successfully derived over Nakagami-m fading channels. Numerical results clearly indicate the strong effects of the residual self-interference (RSI), high frequency, far distances between transmitters and receivers, wide bandwidth, and the altitude of the UAV on the OP and ADR of the MW-FD-UAV-NOMA system. Since the mmWave communications are used, an increase in the frequency, distances, bandwidth, and/or altitude of the UAV will significantly reduce the performance of the MW-FD-UAV-NOMA system. Importantly, when the UAV flies nearer the base station, the performance of the MW-FD-UAV-NOMA system can be better than when the UAV flies nearer the users. Finally, the exactness of the obtained expressions is confirmed via Monte-Carlo simulations.

Optimal scheduling and quantitative analysis for multi-flying warehouse scheduling problem: Amazon airborne fulfillment center

The outstanding advantages of unmanned aerial vehicles (UAVs) have led to an intensive study of the UAV system over the past decade. A variety of potential UAV applications are emerging nowadays, and UAV delivery is one in the limelight. However, one of the critical issues for UAV delivery service is how UAVs can be persistently used within their fundamental limited battery capacity. One recent approach to alleviate the weaknesses is using a cooperative system of UAVs with movable stations for the delivery logistics. The movable stations move across the field of operation, providing persistent delivery service using UAVs by efficiently replenishing the consumables of UAVs. In this study, one of the compelling cooperative delivery services called the flying warehouse system is investigated. The concept of the flying warehouse delivery service has firstly filed as a patent by the e-commerce giant Amazon. The service utilizes airborne fulfillment centers (AFCs), an airship that floats at an altitude of around 45,000. The airship has stocked with inventory, and when a customer places an order, UAVs fly down and deliver the package. The AFC is a novel delivery system that allows delivery in under 10 min and expands Amazon Prime members' access to UAV shipping. On the other hand, AFC is a challenging system that must extract the simultaneous operating schedule of the different system elements. For this purpose, mixed integer linear programming (MILP) is developed. The proposed MILP enables the derivation of service schedules of multiple AFCs and their components. The validity of the MILP is tested through case studies and quantitatively investigated system managerial issues. In addition, we proposed a clustering-based solution approach to provide computational efficiency for real-world case problems.

Weed Detection in Maize Fields by UAV Images Based on Crop Row Preprocessing and Improved YOLOv4

Effective maize and weed detection plays an important role in farmland management, which helps to improve yield and save herbicide resources. Due to their convenience and high resolution, Unmanned Aerial Vehicles (UAVs) are widely used in weed detection. However, there are some challenging problems in weed detection: (i) the cost of labeling is high, the image contains many plants, and annotation of the image is time-consuming and labor-intensive; (ii) the number of maize is much larger than the number of weed in the field, and this imbalance of samples leads to decreased recognition accuracy; and (iii) maize and weed have similar colors, textures, and shapes, which are difficult to identify when an UAV flies at a comparatively high altitude. To solve these problems, we propose a new weed detection framework in this paper. First, to balance the samples and reduce the cost of labeling, a lightweight model YOLOv4-Tiny was exploited to detect and mask the maize rows so that it was only necessary to label weeds on the masked image. Second, the improved YOLOv4 was used as a weed detection model. We introduced the Meta-ACON activation function, added the Convolutional Block Attention Module (CBAM), and replaced the Non-Maximum Suppression (NMS) with Soft Non-Maximum Suppression (Soft-NMS). Moreover, the distributions and counts of weeds were analyzed, which was useful for variable herbicide spraying. The results showed that the total number of labels for 1000 images decrease by half, from 33,572 to 17,126. The improved YOLOv4 had a mean average precision (mAP) of 86.89%.