The paper introduces PIXOR, a real-time 3D object detection system designed for autonomous driving applications. PIXOR addresses the challenge of efficient and accurate 3D object detection from point clouds, which is crucial for safety in autonomous vehicles. The system is a single-stage, proposal-free detector that outputs oriented 3D object estimates from pixel-wise neural network predictions. Key contributions include the use of Bird's Eye View (BEV) representation to reduce computational complexity and improve efficiency, and a fully-convolutional network architecture that balances high accuracy and real-time performance. The model is trained on two datasets: the KITTI BEV object detection benchmark and a large-scale 3D vehicle detection dataset (TOR4D). PIXOR achieves superior performance in terms of Average Precision (AP) compared to state-of-the-art methods while maintaining a frame rate of over 28 FPS. The paper also includes detailed ablation studies and a comparison with other methods, demonstrating the effectiveness and scalability of PIXOR.The paper introduces PIXOR, a real-time 3D object detection system designed for autonomous driving applications. PIXOR addresses the challenge of efficient and accurate 3D object detection from point clouds, which is crucial for safety in autonomous vehicles. The system is a single-stage, proposal-free detector that outputs oriented 3D object estimates from pixel-wise neural network predictions. Key contributions include the use of Bird's Eye View (BEV) representation to reduce computational complexity and improve efficiency, and a fully-convolutional network architecture that balances high accuracy and real-time performance. The model is trained on two datasets: the KITTI BEV object detection benchmark and a large-scale 3D vehicle detection dataset (TOR4D). PIXOR achieves superior performance in terms of Average Precision (AP) compared to state-of-the-art methods while maintaining a frame rate of over 28 FPS. The paper also includes detailed ablation studies and a comparison with other methods, demonstrating the effectiveness and scalability of PIXOR.