Object detection and tracking

Hello, my name is Ioan and I'm currently a student preparing my master degree in computer science. I'm passionate in image processing and computer vision which lead me in building my master dissertation in this field. Having said that, I am working on a project which aims in detecting all the objects, to be specific animals of a certain kind, which are passing on a platform. The project also involves counting these objects as they pass over a reference ROI. The objects can collide on their way through the platform at which point they slightly modify their shapes, the colors are in a certain spectrum going from shades of brown to black and could be affected from the environmental light/sun. Having access to a friend's farm I choose to begin my work there. I have recordings of the animals running over the platform, as well as extracted samples to use as positive detection and started using HAAR/LBP cascades and HOG detection techniques for achieving the project scope. I also experimented using Fast RCNN. At this moment I am currently stuck somewhere on the way and I was looking for help on my project. My main concern is the low accuracy of all tested methods. I would like to discuss with you aspects of my project idea and based on all details I'm interested in a price offer for a prototype. For more information, please contact me. Thank you and kind regards, Ioan