5.4.2 PhotoCity

PhotoCity solves da data inganci da daukan samfur matsaloli a rarraba data collection.

Yanar Gizo kamar Flickr da kuma Facebook taimaka mutane su raba hotuna da abokai da iyali, amma su ma halitta babbar riƙe photos cewa za a iya amfani ga sauran dalilai. Alal misali, Agarwal et al. (2011) ƙoƙarin yin amfani da wadannan photos to "Gina Roma a Day" ta hanyar amfani da 150,000 hotuna na Roma ya halicci 3D maimaitawa na birnin. Domin yawon shakatawa sites kamar Coliseum akwai isa hotuna online don samar 3D model (Figure 5.10), amma ingancin wadannan mujallu da aka iyakance da cewa mafi yawan hotunan da aka dauka daga wannan wurin hutawa ra'ayoyi, da barin rabo daga cikin gine-gine unphotographed. Bugu da ari, ga mafi yawan sassa na birnin, bai isa ba photos kasance available. Saboda haka, ta yin amfani da gano bayanai daga photo repositories, shi ne ba zai yiwu ka sake duk na Roma. Amma, abin da idan masu sa kai za a iya soja kurum marar tattara zama dole photos to hakika "Ku gina Roma, a cikin yini?"

Figure 5.10: A 3-D maimaitawa na Coliseum daga babban tari of 2-D images daga aikin "Building Roma, a cikin yini." The triangles wakilci wurare daga abin da hotunan da aka dauka (Agarwal et al. 2011) .

Domin ba dama da niyya tarin manyan lambobi na photos, Kathleen Tuite da kuma abokan aiki raya PhotoCity, a photo-Ana aikawa game. Daya kyau bangare na PhotoCity shi ne cewa shi ya juya yiwuwar laborious aiki na data tarin-Ana aikawa photos-cikin wani game-kamar aiki shafe teams, gidãje, kuma flags (Figure 5.11). The zane na PhotoCity kuma elegantly solves da daukan samfur da kuma bayanan quality kalubalen eBird da sauran rarraba data tarin ayyukan.

Figure 5.11: PhotoCity juya yiwuwar laborious aiki na tattara bayanai (watau loda photos) da kuma juya shi a cikin wani wasan (Tuite et al. 2011) .

PhotoCity aka fara tura taimaka a 3D maimaitawa biyu jami'o'i: Cornell University kuma Jami'ar Washington. Players a kowane harabar iya duba halin yanzu jihar na maimaitawa model na harabar. Sa'an nan kuma, za su iya sami maki da loda images cewa fadada halin yanzu model. Alal misali, idan na yanzu model na Uris Library (a Cornell) ne ƙwarai Prakiraan, a player iya sami maki da loda sabon hotunan da shi. Kafofin yada, da photos da aka uploaded dole zoba da data kasance photos, sai su za a iya inganta, da kuma adadin da maki a wasan samu dogara ne a kan adadin da cewa su photo ƙara halin yanzu model. A ƙarshe, da masu bincike sun iya yin amfani da wadannan uploaded hotuna don ƙirƙirar high ƙuduri 3D model na gine-gine a kan duka cibiyoyin karatun (Figure 5.12).

Figure 5.12: The PhotoCity game sa masu bincike da kuma mahalarta su haifar high quality-3D model na gine-gine ta yin amfani da hotuna da uploaded by mahalarta (Tuite et al. 2011) .

The zane na PhotoCity elegantly solves biyu matsalolin: data Ingancin kuma daukan samfur. Na farko, photos da aka inganta da matching su da m photos da suke bi da bi dace to previous photos duk hanyar da baya ga iri photos da aka uploaded by bincike. A wasu kalmomin, saboda wannan gina-in redundancy, yana da matukar wuya ga tsarin yarda bad data. Na biyu, da cin kwallo tsarin halitta basu horo mahalarta tattara mafi muhimmanci da-ba mafi m-data. A gaskiya, a nan ne wasu daga cikin dabarun da 'yan wasan ya bayyana amfani domin tsirfatãwa more maki, wanda yake shi ne daidai tattara more m data (Tuite et al. 2011) :

• "[Na yi kokarin] m lokacin da rana da lighting cewa wasu hotuna da aka dauka. wannan zai taimaka wajen hana kin amincewa da wasan. Tare da cewa ya ce, hadari days su ne mafi kyau da nisa a lõkacin da ake rubutu da sasanninta, saboda kasa bambanci taimaki game adadi fitar da lissafi daga hotuna. "

• "A lokacin da ya m, na amfani kyamara ta anti-shake fasali don ba da damar kaina ya dauki photos yayin tafiya a kusa da wani zone. Wannan yarda da ni zuwa sama kintsattse photos, alhãli kuwa ba da ciwon dakatar ta stride. Har ila yau bonus: m mutane stared a gare ni. "

• "Samun yawa hotuna daya gini da 5 megapixel kamara, sa'an nan kuma zuwa gida ya sallama, wani lokacin har zuwa 5 gigs a karshen mako shoot, shi ne na farko photo kama dabarun. Shirya hotuna a waje rumbun kwamfutarka manyan fayiloli zuwa harabar yankin, gina, to, fuskar ginin azurta kyau matsayi ga tsarin uploads. "

Wadannan kalamai daga mahalarta nuna cewa a lokacin da aka azurta su dace feedback, za su iya zama quite gwani a tattara bayanai sha'awa bincike.

Overall, da PhotoCity aikin nuna cewa daukan samfur da kuma bayanan quality ba warkarwa matsaloli a rarraba data collection. Bugu da ari, shi ya nuna cewa rarraba data tarin ayyukan da ake bai iyakance don ayyuka da cewa mutane an riga yin ta wata hanya, irin kallon tsuntsaye. Da dama zane, masu sa kai za a iya karfafa yin wasu abubuwa kuma.