Unmanned Aerial Vehicles (UAVs) have seen significant advancement in recent years and are now widely deployed for applications such as surveillance, remote sensing, disaster management, and environmental monitoring. Their ability to capture and transmit digital images in real time makes them a crucial tool in various industries. However, ensuring the security and authenticity of UAV-transmitted images remains a major challenge due to the risks of unauthorized access, data manipulation, and other cyber threats. Digital image watermarking (DIW) is widely used to ensure the security of digital imagery transmitted through UAV systems. This work proposes a hybrid zero-watermarking framework to enhance the security and robustness of UAV-transmitted images without modifying the original image. Deep learning (DL) is employed to extract high-level image features, enabling an effective and resilient watermarking process. The framework integrates Discrete Wavelet Transform (DWT) for multi-resolution analysis, Pretrained EfficientNet-B0 for feature extraction, and Discrete Cosine Transform (DCT) to encode watermark-related information in the frequency domain. To further strengthen security, a 2D chaotic encryption mechanism generates chaotic key vectors (Sx, Sy), introducing randomness into the watermark structure. The encrypted watermark W enc is embedded in the DWT–DCT-transformed EfficientNet features to produce a Logical Key (LK) for verification and ownership authentication. Experimental results show that the proposed zero-watermarking framework is highly robust against JPEG compression (NC ≥ 0.99 for QF ≥ 50), Median and Gaussian filtering (NC ≥ 0.94), sharpening (NC ≥ 0.99), and Butterworth filtering (NC ≥ 0.99), while maintaining a low Bit Error Rate (BER ≤ 0.005) under various distortions. Furthermore, the keyspace of the 2D chaotic system is 2256, which provides strong resistance against brute-force attacks. Compared with conventional watermarking schemes, the proposed framework preserves image integrity, offers strong authentication, and supports copyright protection, making it suitable for secure UAV image transmission, forensic validation, and digital rights management in aerial imaging applications. Comparative results against recent state-of-the-art schemes confirm that the proposed framework is highly effective for UAV imagery and related applications.