CAFFE model generation

I need generation of CAFFE + MobileNet SSD model based on images (you will get them) so that any OpenCV example app (e.g. its readNetFromCaffe() function) can accurately infer the moving objects on the road into the following four classes:
1. Vehicles (1000 images)
2. Cargo vehicles (1000 images: 700 trucks and 300 vans should go into same class)
3. Citybus (1000 images)
4. Touristbus (12 images)

All images are in 704x576 resolution and are labeled as such.
Inference will run on Win10 machine inside my own C++ app that is based upon OpenCV ver. 4.10 and *from realtime video* . Recognition accuracy should reach >= 95%. FPS rate should be >= 10 FPS (e.g raw inference time <= 100 ms on Intel I7 CPU).

Model might be perhaps done in Caffe2 instead (if you are more convenient with it) but the final result must be fully functional and accurate in the above mentioned environment.
About the recuiter
Member since Nov 11, 2022
James Stan
from Mississippi, United States

---

---

Offer to work on this project closes in 127 days!
Are you interested in this Opportunity?

Looking for help? Checkout our video tutorial
How to search and apply for jobs

How to apply? Do you have more questions about the Job?
See frequently asked questions