Help with ERV analysis

[icd20]

Hi,
Very new to this! Apologies if not enough information given.
I'm trying to count the change in ERV expression before and after treatment of human cell lines with a drug. I have been using the gEVE database, which provides a GTF annotation for human genomes (http://geve.med.u-tokai.ac.jp/download/).
My pipeline is: generate genome indices using STAR --> trim reads (fasta files) using trimmomatic --> sort, remove blacklisted regions --> align trimmed reads to genome using STAR again --> count reads using featureCounts.

All seems fine except counting reads: I get a very low percentage of assigned reads (usually 0.1%) and these are mainly due to "Unassigned_NoFeatures". This is my input for featureCounts:
fc_ERV <- featureCounts(files=filenames, annot.ext="Hsap38.geve.v1.gtf", isGTFAnnotationFile=TRUE, GTF.featureType="CDS", countMultiMappingReads=TRUE, genome='Homo_sapiens.GRCh38.dna.primary_assembly.fa', isPairedEnd=TRUE,nthread=20)
I have used "GTF.featureType="CDS"" because the GTF from gEVE does not have an exon column, only CDS.
Any ideas on what I'm doing wrong? I am new to Bioinformatics, any help would be much appreciated.