The main contents of the project include traffic sign detection and recognition,
accelerating the detection process using visual features, traffic signal recognition
and simulation, lane line and feasible area detection and simulation. Based on the
improved faster RCNN and threshold segmentation, traffic sign recognition and visual
feature acceleration are realized.
The main content of the project is the data-driven intelligent cockpit. Based on the
open source platform airsim, the project uses high-precision map to build a static
virtual scene, builds a dynamic scene in the virtual scene, and uses the cockpit
sensor simulation and vehicle dynamics simulation to realize the perception of
traffic environment.
The main content of the project is fine-grained vehicle recognition. Vehicle
recognition is realized based on BCNN and improved BCNNthe vehicle model can be
identified with high accuracy, and the effect is still very good after model compression.
