Multiple instance learning on deep features for weakly supervised object detection with extreme domain shifts

3 Aug 2020  ·  Nicolas Gonthier, Saïd Ladjal, Yann Gousseau ·

Weakly supervised object detection (WSOD) using only image-level annotations has attracted a growing attention over the past few years. Whereas such task is typically addressed with a domain-specific solution focused on natural images, we show that a simple multiple instance approach applied on pre-trained deep features yields excellent performances on non-photographic datasets, possibly including new classes. The approach does not include any fine-tuning or cross-domain learning and is therefore efficient and possibly applicable to arbitrary datasets and classes. We investigate several flavors of the proposed approach, some including multi-layers perceptron and polyhedral classifiers. Despite its simplicity, our method shows competitive results on a range of publicly available datasets, including paintings (People-Art, IconArt), watercolors, cliparts and comics and allows to quickly learn unseen visual categories.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Weakly Supervised Object Detection CASPAPaintings MI-max Mean mAP 16.2 # 1
Weakly Supervised Object Detection Clipart1k MI-max MAP 38.4 # 7
Weakly Supervised Object Detection Comic2k MI-max MAP 27 # 8
Weakly Supervised Object Detection IconArt MI_Net [wang_revisiting_2018] MAP 15.1 # 1
Weakly Supervised Object Detection PeopleArt Polyhedral MI-max MAP 58.3 # 1
Weakly Supervised Object Detection Watercolor2k MI-max MAP 49.5 # 11


No methods listed for this paper. Add relevant methods here