The target edits the same video with a blue current pedestrian, a yellow future appearance cropped from real future frames, and footprints connecting current and future positions.
Edit the input pedestrian video to visualize where the tracked pedestrian will be in the near future. Use only visual overlays grounded in the future tracking annotations: keep the current pedestrian highlighted with a blue bounding box in every target frame, show the same pedestrian as a yellow future appearance at the annotated future position, and connect the current and future foot positions with visual footprints. Do not rely on explanatory text inside the video.
{
"schema": "v2v-metadata/2.0",
"task": {
"id": "Task_108_pedestrian_future_position_edit",
"type": "v2v",
"capability": "future prediction and video editing",
"domain": "transportation / pedestrian scenes"
},
"source": {
"dataset": "MOT17 / MOTChallenge pedestrian tracking benchmark",
"dataset_url": "https://huggingface.co/datasets/Lekim89/mot17-parquet",
"license": "cc-by-nc-sa-4.0",
"split": "train",
"sequence": "MOT17-09",
"track_id": 12,
"raw_frame_start": 220,
"raw_frame_count": 61,
"rendered_input_frame_count": 41,
"future_horizon_frames": 20,
"raw_frame_pattern": "train/MOT17-09/img1/%06d.jpg",
"groundtruth_file": "train/gt_train.parquet"
},
"rendering": {
"fps": 10,
"width": 640,
"height": 360,
"current_pedestrian": "blue outline around current true bbox",
"future_pedestrian": "yellow-tinted crop of the same pedestrian from true future frame bbox",
"connector": "yellow footprints between current and future foot positions",
"text_in_video": false,
"generated_person_pixels": false
},
"provenance": {
"generator": "New_Tasks/Task_108_pedestrian_future_position_edit/scripts/generate.py",
"config": "New_Tasks/Task_108_pedestrian_future_position_edit/config/sample_000000000.json",
"deterministic": true
}
}