Research and development

Multimodal perception for robotics in agricultural and forestry contexts

Ripping depth perception and VRT fertilization

TBA

Visible and thermal forest tree trunk detection

Detection of tree trunks in forest environment using different vision-based sensors (GoPro Hero6, FLIR M232, ZED Stereo, Allied Mako G-125, OAK-D) and several Deep Learning object detection models (SSD MobileNet, SSD Inception, SSD ResNet, SSDLite MobileDet, EfficientDet Lite, YOLOv4, YOLOv5, YOLOv7, YOLOR and DETR-ResNet) running inference on distinct hardware platforms - Raspberry Pi 4B (CPU), NVIDIA Jetson Nano (GPU), Google Coral USB Accelerator (TPU) and OAK-D's Intel Movidius Myriad X (VPU) - following an Edge-AI approach.

Visible tree trunk detection here!

Thermal tree trunk detection here!

Multispectral perception of forest tree trunk types (genus) using YOLOv5 and YOLOv8 models

Detection and segmentation of two forest tree trunk types - Eucalyptus and Pine - using YOLOv5 and YOLOv8 Deep Learning models and considering 4-channel multispectral images (R,G,B,NIR).

Github repository of YOLOv5 for training and testing on 4-channel images: https://github.com/daniqsilva25/yolov5_4chan

Github repository of YOLOv8 for training and testing on 4-channel images: https://github.com/daniqsilva25/ultralytics_4chan

Automatic eucalyptus stem selection algorithm for thinning

TBA

Watch it here!

Vegetation line clearing perception system and implement control

TBA

Watch it here!