RoogleVision icon indicating copy to clipboard operation
RoogleVision copied to clipboard

Request Status Code: 502

Open claudemeisch opened this issue 8 years ago • 1 comments

Hey, First of all, thank you for the package. I try to classify landmarks with the google vision API on Flickr-data. I do not want to download the data but try to classify the files by passing URLs. But I get Request Status Code: 502. Some error with a JSON content? It says no JSON content found. I cannot find any help on this? Any idea how the error is produced?

On windows x86-64, R version 3.4.1 (2017-04-21) In Detail i do: require(RoogleVision) require(googleAuthR) require(magrittr)

options("googleAuthR.client_id" = ....) options("googleAuthR.client_secret" = ....) options("googleAuthR.scopes.selected" = c("https://www.googleapis.com/auth/cloud-platform")) googleAuthR::gar_auth()

Then i load a datatable with URL's of Flickr images which I want to classifiy. For this, i first read the datatable with Flickr <- data.table::fread (Flickr_data.csv) and then go

getGoogleVisionResponse(imagePath = Flickr$URL[i], feature="LANDMARK_DETECTION")

with Flickr$URL[i] being characters

Request Status Code: 502 Error : lexical error: invalid char in json text. smallerthansign!DOCTYPE html> <html lang=engreaterthansign (right here) ------^

Trying again: 1 of 5 Trying again: 2 of 5 Trying again: 3 of 5 Trying again: 4 of 5 Trying again: 5 of 5 Warning message: No JSON content found in request

opened an issue here already but got referred to this site :) <- https://github.com/MarkEdmondson1234/googleAuthR/issues/80#issuecomment-324619002

Thank you!

claudemeisch avatar Sep 04 '17 09:09 claudemeisch

Hi, it seems that your request url does not provide a path to an image. Have you tried the call: getGoogleVisionResponse("https://media-cdn.tripadvisor.com/media/photo-s/02/6b/c2/19/filename-48842881-jpg.jpg", feature="LANDMARK_DETECTION")

Does that one work? Do you get a proper response by manually setting the path? getGoogleVisionResponse(imagePath = "url.path", feature="LANDMARK_DETECTION")

flovv avatar Sep 05 '17 20:09 flovv