Uploaded September 2015 | Updated September 2026, 4 hours ago
Pedestrian Detectors applied in challenging offroad environment for different false positive rates across our full test set.
Each example shows bounding boxes for detections from ACF (green), DPM (red), and CNN (blue), along with the labeled ground truth (white).
The left image in each pair uses a threshold that corresponds to an average false positive rate of 0.01 per image and the right image corresponds to 0.1 per image over this dataset.
For more info, please visit:
bit.ly/1MbWQUD
T. Tabor, Z. Pezzementi, C. Vallespi and C. Wellington, 'People in the Weeds: Pedestrian Detection Goes Off-road', in 2015 IEEE International Symposium on Safety, Security, and Rescue Robotics, Purdue University, West Lafayette, IN, 2015.
Robotics offers a great opportunity to improve efficiency while also improving safety, but reliable detection of humans in off-road environments remains a key challenge. We present a person detector evaluation on a dataset collected from an autonomous tractor in an orchard environment representing challenging conditions with significant occlusion from weeds and branches as well as non-standing poses. We apply three image-only algorithms from urban pedestrian detection to better understand how well these approaches work in this domain. We evaluate the Aggregate Channel Features (ACF) and Deformable Parts Model (DPM) algorithms from the literature, as well as our own implementation of a Convolutional Neural Network (CNN). We show that the traditional performance metric used in the pedestrian detection literature is extremely sensitive to parameterization. When applied in domains like this one, where localization is challenging due to high background texture and occlusion, the choice of overlap threshold strongly affects measured performance. Using a permissive overlap threshold, we found that ACF, DPM, and CNN perform similarly overall in this domain, although they each have different failure modes.
Pedestrian Detectors applied in challenging offroad environment for different false positive rates across our full test set.
Each example shows bounding boxes for detections from ACF (green), DPM (red), and CNN (blue), along with the labeled ground truth (white).
The left image in each pair uses a threshold that corresponds to an average false positive rate of 0.01 per image and the right image corresponds to 0.1 per image over this dataset.
For more info, please visit:
bit.ly/1MbWQUD
T. Tabor, Z. Pezzementi, C. Vallespi and C. Wellington, 'People in the Weeds: Pedestrian Detection Goes Off-road', in 2015 IEEE International Symposium on Safety, Security, and Rescue Robotics, Purdue University, West Lafayette, IN, 2015.
Robotics offers a great opportunity to improve efficiency while also improving safety, but reliable detection of humans in off-road environments remains a key challenge. We present a person detector evaluation on a dataset collected from an autonomous tractor in an orchard environment representing challenging conditions with significant occlusion from weeds and branches as well as non-standing poses. We apply three image-only algorithms from urban pedestrian detection to better understand how well these approaches work in this domain. We evaluate the Aggregate Channel Features (ACF) and Deformable Parts Model (DPM) algorithms from the literature, as well as our own implementation of a Convolutional Neural Network (CNN). We show that the traditional performance metric used in the pedestrian detection literature is extremely sensitive to parameterization. When applied in domains like this one, where localization is challenging due to high background texture and occlusion, the choice of overlap threshold strongly affects measured performance. Using a permissive overlap threshold, we found that ACF, DPM, and CNN perform similarly overall in this domain, although they each have different failure modes.
