CompleteTheMap V

Hi everyone,

I’m posting here to let you know about the latest CompleteTheMap that we are running, as announced on the blog. As usual, we’re encouraging people to submit locations and contribute as much imagery within those locations as they can during the challenge period. Contributions will be scored with the top 3 finalists being awarded some pretty fantastic cameras. The focus this time is quality! More on that below.

Submit a location.

To show up on the leaderboard, you need to be contributing imagery in one the challenge locations. Challenge locations can be submitted by anyone, but each person is only allowed to submit one challenge location. We will accept location submissions until May 28th.

Here are the locations nominated so far.

Looking more closely at North America.

Looking more closely at Ukraine.

Scoring changes

score(s) = quality(weight(density(s)) * UKM(s))

The scoring system is going to be the same as the last challenge, but with one key difference. A quality weighting will be applied manually to each score. This quality weighting will be multiplied on top of your regular score, but everyone will have a quality weighting of 1 by default. We’ll have a look at the average quality of your sequences and apply a rating between 0 and 10. Your score will then be multiplied by the weighting, with the weighting being your (quality rating/10). In other words, a number between 0 and 1.

So if you finish the challenge with a score of 120,000 and a quality score of 6, your score would be multiplied by 0.6, giving you a final score of 72,000.

There is no automatic way to apply this score currently, so we’re going to be doing it manually. There is of course a level of subjectivity here, but the following sequences should give you an idea of what we’re looking for. Your quality score can change throughout the challenge if the sequences you contribute get better or worse.


10: Driving sequence. Taken with smartphone.


8: Driving sequence. Dash cam time stamp in the bottom right, small portion of the car visible in the image.


6: Walking sequence. Inconsistent compass angle and compass not level with the horizon.


4: Driving sequence. Camera at an angle, lighting poor, and very grainy.


2: Driving sequence. Rain on the windshield, reflections, dash cam info in the top left, and a significant portion of the dashboard obstructing the view.


0: Driving sequence. Night sequence with camera at an angle and very blurry.

As in previous challenges, UKM can take a day or so to update, so be patient after upload and you will eventually see it update accordingly if you were the first person in the challenge to capture that given road/track.

#CompleteTheMap V Timeline

Prizes

Prizes have proved popular and well-utilized in previous challenges, so we’re continuing this with some excellent cameras available for those who finish in the top three.

1st place : GoPro Fusion 360º + two 128GB SD cards
2nd place : GoPro Hero 7 Black + 128GB card
3rd place : Blackvue DR900s-1ch dashcam + 128GB card

1st and 2nd place will have the option to choose the 2nd or 3rd prize should they prefer that option. The winners will be announced in early July once we have had time to confirm all scores have been legitimately obtained and that all images are uploaded.

Leaderboard

When the leaderboard goes live, you’ll be able to view all locations and scores from here. Be patient, it’s loading!

Thanks for making CompleteTheMap a regular part of the Mapillary calendar.

Ed & the Mapillary team

P.S. We changed it to #CompleteTheMap because it’s much easier to read.

4 Likes

[start 360 image promo]
Expected a 360 image to get 10 points, and a good smartphone picture 8 (or 9)
Or a “360 bonus” of 2 points across the board?

(Wouldn’t such pictures eventually become the most ideal for Mapillary?)

PS: good to see that first place gets a 360 cam!
[/start 360 image promo]

Yes they’re most ideal, but most people don’t have access to those cameras yet. We’re looking at ways to factor that into the score for future challenges, but need to do it in a way that is still fair to those doing the best they can with their phone.

Good to read that you see 360 images as most ideal.

You’ve added a “value rating” for the quality of the images. That should mean 360 images are better then standard. So I think the “360 bonus” is the best way to go?
My 2 cents: set the best “standard image” on 9 points and add a one point bonus across the board for a 360 image. Excluding a really blurry 360 image. in 360 they are as good as useless also, so no points for my picture there :stuck_out_tongue:

360 --> 2x

??

IF
Camera make & model = Blackvue DR900S-1CH = ( Mapillary – complete-the-map-iv – 3rd place)
Image resolution = 3840 x 2160 = 4K
light on the glass = 0 = none
horizon = 0 degree
vertical = 0 degree
crop = TRUE

if I load the crop, I will have a multiplier = 1.0
but if I load the original, I will have a multiplier = 0.7

all right?

sample ORIGINAL:

and…

Or just play the game for the sake of improving coverage and enjoying the process, not trying to beat the system. Geez.

6 Likes

This is something that we will look at in future challenges. We need to add a 360º true/false call in the leaderboard API that we use of this.

Part of the challenge is also encouraging new contributors who may be using their phone. As the winner receives a 360º camera, we run the risk of having the same person win each time because they were the one that received the 360º camera in a previous challenge.

So it’s a bit of a balancing act, but we all agree 360º images are ideal.

I would say this is similar to an 8. So multiplier of 0.8

We’re working to automate this evaluation, but in the meantime it is a subjective, manual process :wink.

I’d say this is close enough. Obviously the closer to the true horizon the better, but we’re not going to be judging people for a few degrees here and there.

image

SUN - BlackVue900 Edition (autofix) ???

Do you have a link to the uploaded image?

(file storage)


original files
and
100 images as sample

image

link

stat: day - 0
D0

D2

D3

D4

D6

D10 (1/3)

D11

D15

https://mapillary.github.io/mapillary_greenhouse/global-challenge/q2-2019/

Sorry for the delay. Grids are taking a while to generate due to having ~95 locations submitted. Don’t worry as all collected imagery will count and we should have them up soon.

1 Like

Driving sequence. Camera at an angle, lighting poor, and very grainy.

you dont see my 6 “bad” photo as 270 degree panoramic??
(for all sequences)

-90
image
-45 (effect rigth lense)
image
0
image
+45 (effect left lense)
image
+90
image

Hi,

how do I get to know, if my submission was accepted?

Istefanos

https://mapillary.github.io/mapillary_greenhouse/global-challenge/q2-2019/

1 Like