Data Labeling Services by iMerit

We have been working with iMerit on our NLP related labeling tasks. We worked with them for both NER and text classification problems. The labeling task was delivered on time and the quality of the work was very good. We will continue work with them for our future projects. We recommend working with iMerit team if you are looking reliable labeling with your NLP problems.

From a support perspective the staff were friendly, incredibly professional and fast to respond to emails. From a quality point of view, they underestimated the complexity of our work. We had several discussions with them detailing our requirements and even supplied a comprehensive report. They were confident they could complete the jobs in a given time frame. Unfortunately, they took long enough to work through our data that some of our jobs expired. Which resulted in us having to restart our jobs, which incurred additional cost and effort.

I think they would have delivered better results if the work wasn't underestimated and had fewer images per job. Though we followed their advice on the number of images and time constraints.

iMerit has been a great partner to work with. Sentera has found that having a collaborator inside the AWS ecosystem, which is used company-wide, takes care of a lot of quirks that come with data management and format conversions. iMerit representatives communicate the progress of the annotation jobs daily, are open to suggestions, and allocate resources to make sure that our needs are met. We have tested the object annotation workflow for a few different data products now and the collaboration has been time and cost-effective for our company.


My team found out that just blindly assigning a labeling job through the AWS interface will most probably not yield the desired results, especially when the annotation needs are unique with respect to cost, the accuracy of the bounding boxes, etc. As a tip for anyone wishing to start the process with iMerit, I would say contact them first. They are fast to respond and explain how their pipelines can accommodate your needs. With every new data annotation project, you will go through a few iterations with iMerit representatives where you send a sample of images for annotation, you assess the results, make corrections, and repeat until the desired balance between cost and accuracy is met. Overall, our collaboration has been satisfactory for Sentera and we will continue collaborating with the iMerit team.

I have to say that having dealt with iMerit only once gave me already strong rationale to keep on using their services moving forward. Not only they have great customer service but they also started to label a few pictures [following instructions provided] and then came back with an estimated budget for completing the task [per different set of iterations], awaiting customer (my) green light to move on. This was great!, since the complexity of the annotation task may differ job from job. Plus, the quality of labels exceeded that of Mechanical Turk.

An awesome company. Very fast to respond to requests, always available and communicative, and they provide high quality label data at a low cost. Would definitely use them for any future projects. We had them label tens of thousands of images with instance segmentation annotations. Similar numbers for object detection, segmentation and classification. Their results were far far better than what we were getting from the public mechanical turk labelers. I am 100% satisfied.