English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Existing approaches for autonomous control of pan-tilt-zoom (PTZ) cameras usemultiple stages where object detection and localization are performedseparately from the control of the PTZ mechanisms. These approaches requiremanual labels and suffer from performance bottlenecks due to error propagationacross the multi-stage flow of information. The large size of object detectionneural networks also makes prior solutions infeasible for real-time deploymentin resource-constrained devices. We present an end-to-end deep reinforcementlearning (RL) solution called Eagle to train a neural network policy thatdirectly takes images as input to control the PTZ camera. Trainingreinforcement learning is cumbersome in the real world due to labeling effort,runtime environment stochasticity, and fragile experimental setups. Weintroduce a photo-realistic simulation framework for training and evaluation ofPTZ camera control policies. Eagle achieves superior camera control performanceby maintaining the object of interest close to the center of captured images athigh resolution and has up to 17% more tracking duration than thestate-of-the-art. Eagle policies are lightweight (90x fewer parameters thanYolo5s) and can run on embedded camera platforms such as Raspberry PI (33 FPS)and Jetson Nano (38 FPS), facilitating real-time PTZ tracking forresource-constrained environments. With domain randomization, Eagle policiestrained in our simulator can be transferred directly to real-world scenarios.

Eagle: End-to-end Deep Reinforcement Learning based Autonomous Control of PTZ Cameras