MLCAS 2019 Challenge - Sorghum head detection

Organized by koushikn - Current server time: Oct. 22, 2019, 5:45 a.m. UTC
Reward $3,700

First phase

Final test
July 13, 2019, midnight UTC

End

Competition Ends
July 28, 2019, 11:59 p.m. UTC

MLCAS 2019 - Sorghum head detection

We are pleased to announce the MLCAS 2019 - Sorghum head detection challenge. This challenge aims at identifying new machine learning approaches to advance the state-of-the-art in Sorghum head detection from UAV images. The detection of sorghum heads from crop row images will be useful in estimating yield. The train, validation and test sets containing around 1300 images are available to download from the "Files" tab under "Participate" section. More details on the MLCAS 2019 workshop can be found at https://register.extension.iastate.edu/mlcas2019/about. For more details on the dataset, please refer to

  • Ghosal, Sambuddha, et al. "A Weakly Supervised Deep Learning Framework for Sorghum Head Detection and Counting." Plant Phenomics 2019 (2019): 1525874.
  • Guo, Wei, et al. "Aerial Imagery Analysis—Quantifying Appearance and Number of Sorghum Heads for Applications in Breeding and Agronomy." Frontiers in plant science 9 (2018): 1544.

Annotation Format

  • The names of the detection files match their corresponding mages(e.g. "C2-R29-G57-DSC01478.txt" represents the detections of the images: "C2-R29-G57-DSC01478.jpeg").
  • Pixel coordinates are in (x,y) format where x represents the column and y represents the row of the image. The first pixel at top left corner of the image is having coordinates (0,0). The left top (x1,y1) coordinates and right bottom (x2,y2) are used for detection annotaion.
  • In these files each line will be in the following format: <class_name> <left> <top> <right> <bottom>.
  • E.g. "C2-R29-G57-DSC01478.txt":
sorghumHeadyieldTrail 0 507 38 545
sorghumHeadyieldTrail 0 942 20 970
sorghumHeadyieldTrail 0 1262 47 1309
sorghumHeadyieldTrail 2 735 22 759
sorghumHeadyieldTrail 6 704 56 750

Submission Format

  • Create a separate detection text file for each test images
  • The names of the detection files must match their corresponding test images(e.g. "C2-R29-G57-DSC01478.txt" represents the detections of the images: "C2-R29-G57-DSC01478.jpeg").
  • In these files each line should be in the following format: <class_name> <confidence> <left> <top> <right> <bottom>).
  • E.g. "C2-R29-G57-DSC01478.txt":
sorghumHeadyieldTrail 0.621189 1 2 20 27
sorghumHeadyieldTrail 0.400580 1 173 20 196
sorghumHeadyieldTrail 0.396825 1 256 23 280
sorghumHeadyieldTrail 0.922975 2 136 16 158
sorghumHeadyieldTrail 0.734653 3 294 18 314
  • All the text files have to be directly compressed into a zip folder for submission. Note: There should be no sub-folder within the submitted zip file.

Dates

June 10, 2019    MLCAS 2019 – Sorghum head detection challenge launched. 
July 28, 2019   Submission deadline at 23:59:59 UTC.
August 5, 2019   Winners announced.
September 11, 2019   Awards presented at the MLCAS 2019 workshop.
September 12, 2019   Winners present their work at the MLCAS 2019 workshop.

 

Prize

The top three winning teams will receive travel grants as listed below (in addition to waiver of registration fees). Students and early career researchers (within five years of receiving their terminal degree) are eligible for the awards. Note that the amounts listed below are for the whole team, not individual participants. Also, in order to be eligible for the award, at least one team member needs to attend the award ceremony on September 11 and present the team’s work on September 12 at MLCAS 2019.

1st prize    $1500 
2nd prize   $1200
3rd prize   $1000 

Organizing Committee

Wei Guo, University of Tokyo, Japan
Soumik Sarkar, Iowa State University, USA
Baskar Ganapathysubramanian, Iowa State University, USA
Asheesh Singh, Iowa State University, USA
Arti Singh, Iowa State University, USA

For questions related to competition, please post on the discussion forum or contact  koushikn@iastate.edu.

Evaluation Criteria

The mean average precision (MAP) with an Instersection Over Union (IOU) of 0.5 is used for evaluating the performance of the object detector.

Terms and Conditions

The annotations in this dataset belong to the organizers of the challenge and is released for non-commercial research purpose only.

Final test

Start: July 13, 2019, midnight

Competition Ends

July 28, 2019, 11:59 p.m.

You must be logged in to participate in competitions.

Sign In