E-M-S: Event-Monocular-Spacecraft dataset for 6D pose estimation of non-cooperative targets.

This dataset was used in the paper "Cross-Modal Fusion of Monocular images and Neuromorphic streams for 6D Pose Estimation of Non-Cooperative Targets", for studying 6D pose estimation of non-cooperative spacecrafts.

Research Background: Robust pose estimation for non-cooperative space targets is a critical technology for on-orbit servicing and active debris removal. Learning-based monocular vision methods have been widely adopted for spacecraft pose estimation tasks, achieving superior performance compared to traditional approaches. However, extreme illumination conditions found in space pose challenges for these approaches. Unlike traditional cameras, event cameras capture asynchronous brightness changes at the pixel level. These cameras exhibit remarkable attributes, including high dynamic range, low latency, and resistance to motion blur, making them well-suited for high dynamic range scenes and high-speed motion. These features make event cameras an ideal complement to standard RGB cameras for object pose estimation. Combining these two sensor types to leverage their complementary strengths is a promising research endeavor. Currently, there are only a few datasets of event images for spacecraft, and it is important to generate and make public new spacecraft event datasets.

In our paper, we describe the methodology used to generate the dataset. For more information, please refer to our paper:“Cross-Modal Fusion of Monocular images and Neuromorphic streams for 6D Pose  Estimation of Non-Cooperative Targets”

This dataset contains a total of 57200 event.csv  (57,200 event frames), corresponding 57,200 RGB images, 572  RGB videos and ground truth 6D pose labels. The image resolution is 640×480, including different lighting conditions such as normal and low light. The images in the RGB/img folder have the same pose label as the corresponding numbered csv event stream in the Even/event_stream folder. The data format is as follows: 
E-M-S.zip
   --training
      --RGB
         --img{*.jpg}
         --pose {*.txt: Pose format <r1, r2, r3, r4, r5, r6, r7, r8, r9, tx, ty, tz>, the first 9 values represent the rotation matrix, and the last three values represent the translation vector} 
      --Event
         --event_stream{*.csv, event stream with the format<timestamp,x,y,polarity>}
         --pose {*.txt: Pose format <r1, r2, r3, r4, r5, r6, r7, r8, r9, tx, ty, tz>, the first 9 values represent the rotation matrix, and the last three values represent the translation vector}

   --validation (Same data structure as training) 
      --RGB
      --Event

   --testing (Same data structure as training) 
      --RGB
      --Event