I would like to hire a Programmer

The Budget I have initially selected is just for a POC (Proof-of-Concept). Once I am confident the programmer can provide the services I am looking for then the budget will increase. I would like a program developed that can do object recognition on hi-resolution images through Computer Vision using Open CV. The image may have multiple images stitched together known as a “stitched mosaic”. When the object is identified it needs to be counted. For example, if the object is identified as a car we want to count all the cars in the image. The program needs to have a training mode so the more images that are input it will learn what a “car” would look like. The first release of the program would be for cars but future releases will include other object such as manhole covers, cattle, etc. but I don’t anticipate that it will be necessary to count more than 1 ‘type’ of object at a time. We will want to eventually have an Administrator mode where it can be configured what the object the program should be looking for. We will continue to add different types of images in future releases where a selection can be made in the Administrator mode what object should be identified in that instance of the program,. The goal is to first get a ‘car counting’ program and then go from there. The strategy is to eventually provide this as a licensed SAAS (Software as a Service) model hosted in the cloud. We also want to own the source code which would allow us to the development in-house if necessary. We are looking for an approximate quote for this work and the budget I post is negotiable for the right developer. The specific requirements below are for the first release. NOTE: We would like to see a sample of your work of something similar to so we know you have the expertise to do what we are requesting. We look forward to hearing from you. Release 1 Requirements 1. Identify an object and return the count found a. For example, Cars parked in any direction b. The cars would be unobstructed (i.e. blocked by a tree) but if it is partially obstructed it would be a bonus that the car was counted 2. Input: Image or stitched image that are scanned and then a complete count is returned of that object. 3. There should be the option to either have the whole picture scanned automatically or the ability to for the operator to zoom in and select an area for the objects to be counted. 4. Allow for areas to be selected that should not be part of the scan or count. 5. Allow manual selection/deselection that would be added/subtracted from the count a. This would allow for inaccuracies in the program where it didn’t identify an object or if it was obstructed. 6. We would expect that the accuracy of identifying an object would increase based on the “Training” feature of the program. a. We can provide hundreds of images to help in the “Training” mode but that is something that our team can do as well, 7. In future releases we will want to be able to recognize the color of a car to know if that car has been move to determine parking patterns for parking studies. 8. The User Interface is not defined but looking to developer to provide options for a simple User Interface that would eventually include a selection of the Object that is being counted. 9. Reports: Undefined at this point but will need to determine how to produce a report that will summarize the results of the count including date and time of the count from multiple images. 10. Initially this first release should be Windows based but we will eventually move to iOS and mobile but ultimately we want it cloud based to license under a SaaS (Software as a Service) Model. If this can be done with the first release that would be ideal. SAMPLE IMAGED UPLOADED