Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

High-Speed Detector For Low-Powered Devices In Aerial Grasping

Ashish Kumar, Laxmidhar Behera

TL;DR

Fast Fruit Detector (FFD), a resource-efficient, single-stage, and postprocessing-free object detector based on the authors' novel latent object representation module, query assignment, and prediction strategy, is presented.

Abstract

Autonomous aerial harvesting is a highly complex problem because it requires numerous interdisciplinary algorithms to be executed on mini low-powered computing devices. Object detection is one such algorithm that is compute-hungry. In this context, we make the following contributions: (i) Fast Fruit Detector (FFD), a resource-efficient, single-stage, and postprocessing-free object detector based on our novel latent object representation (LOR) module, query assignment, and prediction strategy. FFD achieves 100FPS@FP32 precision on the latest 10W NVIDIA Jetson-NX embedded device while co-existing with other time-critical sub-systems such as control, grasping, SLAM, a major achievement of this work. (ii) a method to generate vast amounts of training data without exhaustive manual labelling of fruit images since they consist of a large number of instances, which increases the labelling cost and time. (iii) an open-source fruit detection dataset having plenty of very small-sized instances that are difficult to detect. Our exhaustive evaluations on our and MinneApple dataset show that FFD, being only a single-scale detector, is more accurate than many representative detectors, e.g. FFD is better than single-scale Faster-RCNN by 10.7AP, multi-scale Faster-RCNN by 2.3AP, and better than latest single-scale YOLO-v8 by 8AP and multi-scale YOLO-v8 by 0.3 while being considerably faster.

High-Speed Detector For Low-Powered Devices In Aerial Grasping

TL;DR

Fast Fruit Detector (FFD), a resource-efficient, single-stage, and postprocessing-free object detector based on the authors' novel latent object representation module, query assignment, and prediction strategy, is presented.

Abstract

Autonomous aerial harvesting is a highly complex problem because it requires numerous interdisciplinary algorithms to be executed on mini low-powered computing devices. Object detection is one such algorithm that is compute-hungry. In this context, we make the following contributions: (i) Fast Fruit Detector (FFD), a resource-efficient, single-stage, and postprocessing-free object detector based on our novel latent object representation (LOR) module, query assignment, and prediction strategy. FFD achieves 100FPS@FP32 precision on the latest 10W NVIDIA Jetson-NX embedded device while co-existing with other time-critical sub-systems such as control, grasping, SLAM, a major achievement of this work. (ii) a method to generate vast amounts of training data without exhaustive manual labelling of fruit images since they consist of a large number of instances, which increases the labelling cost and time. (iii) an open-source fruit detection dataset having plenty of very small-sized instances that are difficult to detect. Our exhaustive evaluations on our and MinneApple dataset show that FFD, being only a single-scale detector, is more accurate than many representative detectors, e.g. FFD is better than single-scale Faster-RCNN by 10.7AP, multi-scale Faster-RCNN by 2.3AP, and better than latest single-scale YOLO-v8 by 8AP and multi-scale YOLO-v8 by 0.3 while being considerably faster.
Paper Structure (39 sections, 11 equations, 9 figures, 8 tables)

This paper contains 39 sections, 11 equations, 9 figures, 8 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Convolutional Neural Network Based Detection
  4. Transformer Based Object Detector
  5. Detection For Fruit Harvesting
  6. Fast Fruit Detector
  7. Backbone
  8. Latent Object Representation (LOR)
  9. Query Transformation QT
  10. Cross Channel Global Context (CCGC)
  11. Delineation
  12. Prediction
  13. Query Assignment
  14. Tiled Hungarian Matching
  15. Objective Function
  16. ...and 24 more sections

Figures (9)

  • Figure 1: Top: Our aerial grasping system for fruit harvesting, and outdoor detection. Bottom: FFD has low training time, high inference speed, and high detection accuracy compared to existing detectors.
  • Figure 2: (a) Faster-RCNN, (b) SSD, (c) DETR, and (d) FFD.
  • Figure 3: Fast-Fruit-Detector (FFD). "GP": Global Pooling,'E': Expand, 'S': Squeeze, and '' Broadcast multiplication.
  • Figure 4: FFD has novel query assignment. In traditional detectors fasterrcnnssd, a query is simply is an anchor. '$t_{ij}$' denotes a tile in the image.
  • Figure 5: Differences between FFD and DETR-like methods.
  • ...and 4 more figures